U.S. patent number 9,946,571 [Application Number 14/573,621] was granted by the patent office on 2018-04-17 for predictive power management in a wireless sensor network using activity costs.
This patent grant is currently assigned to Invent.ly, LLC. The grantee listed for this patent is Invent.ly, LLC. Invention is credited to Stephen J. Brown, Timothy P. Flood, Hector H. Gonzalez-Banos, Clive A. Hallatt, Holden D. Jessup, Daylyn M. Meade.
United States Patent |
9,946,571 |
Brown , et al. |
April 17, 2018 |
Predictive power management in a wireless sensor network using
activity costs
Abstract
A system comprising a plurality of self-powered devices and at
least one remote device. The plurality of self-powered devices may
be configured to (i) perform one or more tasks and (ii) select one
of a plurality of modes of operation. The remote device may be
configured to (a) determine scheduling data for one or more
activities based on a resource capacity of the self-powered devices
and (b) communicate with the self-powered devices. The activities
may comprise one or more of the tasks. The self-powered devices may
determine a computational cost of performing the tasks.
Inventors: |
Brown; Stephen J. (Woodside,
CA), Meade; Daylyn M. (Sebastopol, CA), Flood; Timothy
P. (Sebastopol, CA), Hallatt; Clive A. (Palo Alto,
CA), Jessup; Holden D. (Palo Alto, CA), Gonzalez-Banos;
Hector H. (Mountain View, CA) |
Applicant: |
Name |
City |
State |
Country |
Type |
Invent.ly, LLC |
Woodside |
CA |
US |
|
|
Assignee: |
Invent.ly, LLC (Woodside,
CA)
|
Family
ID: |
61872544 |
Appl.
No.: |
14/573,621 |
Filed: |
December 17, 2014 |
Related U.S. Patent Documents
|
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
Issue Date |
|
|
14169464 |
Jan 31, 2014 |
|
|
|
|
14103209 |
Dec 11, 2013 |
|
|
|
|
13946414 |
Sep 30, 2014 |
8850242 |
|
|
|
13406469 |
Aug 20, 2013 |
8516279 |
|
|
|
12472327 |
Feb 28, 2012 |
8127158 |
|
|
|
11443668 |
May 26, 2009 |
7539882 |
|
|
|
14090099 |
Nov 26, 2013 |
9721210 |
|
|
|
60685976 |
May 30, 2005 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
G06F
9/4893 (20130101); Y02D 10/24 (20180101) |
Current International
Class: |
G06F
9/50 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
Jessup, U.S. Appl. No. 12/472,327, Response, dated Nov. 3, 2011,
7pgs. cited by applicant .
Jessup, U.S. Appl. No. 12/472,327, Notice of Allowance, dated Oct.
20, 2011, 7 pgs. cited by applicant .
Jessup, U.S. Appl. No. 12/472,327, Office Action, dated Mar. 17,
2011, 12 pgs. cited by applicant .
Jessup, U.S. Appl. No. 12/472,327, Response with Terminal
Disclaimer, dated Jul. 8, 2011, 12 pgs. cited by applicant .
Jessup, Holden re U.S. Appl. No. 12/472,327 filed May 26, 2009 re
Amendment dated Nov. 3, 2011. 7 pages. cited by applicant .
Jessup, Holden re U.S. Appl. No. 12/472,327, filed May 26, 2009 re
Notice of Allowance and Fee(s) dated Oct. 20, 2011. 7 pages. cited
by applicant .
Jessup, Holden, U.S. Appl. No. 12/472,327, filed May 26, 2009,
Office Action dated Mar. 17, 2011. 12 pages. cited by applicant
.
Response dated Jul. 8, 2011 to the Office Action dated Mar. 17,
2011 re U.S. Appl. No. 12/472,327 includes Terminal Disclaimer. 12
pages. cited by applicant.
|
Primary Examiner: Yen; Paul J
Attorney, Agent or Firm: Christopher P. Maiorana, PC
Claims
The invention claimed is:
1. A system comprising: a plurality of self-powered devices,
wherein said self-powered devices are configured to (i) perform one
or more tasks and (ii) select one of a plurality of modes of
operation; and at least one remote device configured to (a)
determine scheduling data for one or more activities based on a
resource capacity of said self-powered devices and (b) communicate
with said plurality of self-powered devices, wherein (i) said
activities comprise one or more of said tasks, (ii) said
self-powered devices determine a computational cost of performing
said tasks, (iii) a subset of said self-powered devices comprise
redundant self-powered devices, (iv) said redundant self-powered
devices are configured to implement similar tasks, (v) said
redundant self-powered devices are implemented having different
fixed-costs for performing said similar tasks based on said
selected mode of operation and (vi) said remote device is further
configured to determine said scheduling data based on (a) said
different fixed-costs of said redundant self-powered devices and
(b) said computational cost of implementing said similar tasks on
said redundant self-powered devices.
2. The system according to claim 1, wherein (i) said resource
capacity of said self-powered devices comprises a battery capacity
and (ii) said computational cost is mapped based on a battery
discharge.
3. The system according to claim 1, wherein (i) said self-powered
devices are further configured to store an instruction set
comprising one or more native instructions and (ii) said tasks
comprise a sequence of said native instructions.
4. The system according to claim 3, wherein said instruction set is
updatable.
5. The system according to claim 1, wherein said computational cost
comprises a fixed-cost component based on said selected mode of
operation of said self-powered device.
6. The system according to claim 1, wherein said activities
comprise sending messages to said self-powered devices, updating
said self-powered devices and running diagnostics on said
self-powered devices.
7. The system according to claim 1, wherein (i) said resource
capacity is based on characteristics of said self-powered devices,
(ii) said remote device is further configured to determine
precedence relationships for said activities and (iii) said
precedence relationships are used to determine said scheduling
data.
8. The system according to claim 1, wherein said self-powered
devices are configured to (i) aggregate tasks and (ii) execute said
aggregated tasks together.
9. The system according to claim 8, wherein said aggregated tasks
comprise said tasks determined to be non-urgent tasks.
10. The system according to claim 9, wherein said non-urgent tasks
are determined by said remote device based on context
information.
11. The system according to claim 1, wherein (i) one or more of
said tasks comprise critical tasks and (ii) said self-powered
devices are configured to perform said critical tasks at any
time.
12. The system according to claim 11, wherein a subset of said
resource capacity of said self-powered devices is a reserve charge
to ensure a performance of said critical tasks.
13. The system according to claim 12, wherein said self-powered
device is considered unavailable by said remote device if said
reserve charge is below a threshold for performing said critical
tasks.
14. The system according to claim 12, wherein (i) said remote
device is further configured to update a number of said critical
tasks performed by said self-powered devices, (ii) said reserve
charge of said self-powered devices is increased if said number of
said critical tasks is increased and (iii) said reserve charge of
said self-powered devices is decreased if said number of said
critical tasks is decreased.
15. The system according to claim 12, wherein said remote device is
further configured to (i) remove said reserve charge from said
resource capacity of said self-powered devices and (ii) perform
refinement rounds to update said scheduling data.
16. The system according to claim 1, wherein said remote device is
further configured to form a network of said redundant self-powered
devices from said subset of said self-powered devices by updating
an instruction set of said redundant self-powered devices.
17. The system according to claim 1, wherein (i) said redundant
self-powered devices are implemented having a fixed-cost value of
zero for performing said similar tasks and (ii) said remote device
is further configured to determine said scheduling data based on
(a) said computational cost of implementing said similar tasks on
said redundant self-powered devices and (b) an availability of said
redundant self-powered devices.
18. The system according to claim 1, wherein said redundant
self-powered devices are implemented as a network of sensors.
19. A system comprising: a plurality of self-powered devices,
wherein said self-powered devices are configured to (i) perform one
or more tasks and (ii) select one of a plurality of modes of
operation; and at least one remote device configured to (a)
determine scheduling data for one or more activities based on a
resource capacity of said self-powered devices and (b) communicate
with said plurality of self-powered devices, wherein (i) said
activities comprise one or more of said tasks, (ii) said
self-powered devices determine a computational cost of performing
said tasks, (iii) one or more of said tasks comprise critical
tasks, (iv) said self-powered devices are configured to perform
said critical tasks at any time, (v) a subset of said resource
capacity of said self-powered devices is a reserve charge to ensure
a performance of said critical tasks, (vi) said remote device is
further configured to update a number of said critical tasks
performed by said self-powered devices, (vii) said reserve charge
of said self-powered devices is increased if said number of said
critical tasks is increased and (viii) said reserve charge of said
self-powered devices is decreased if said number of said critical
tasks is decreased.
Description
This application relates to U.S. Ser. No. 14/169,464, filed Jan.
31, 2014, which relates to U.S. Ser. No. 14/103,209, filed Dec. 11,
2013, which relates to U.S. Ser. No. 13/946,414, filed Jul. 19,
2013, now U.S. Pat. No. 8,850,242, which relates to U.S. Ser. No.
13/406,469, filed Feb. 27, 2012, now U.S. Pat. No. 8,516,279, which
relates to U.S. Ser. No. 12/472,327, filed May 26, 2009, now U.S.
Pat. No. 8,127,158, which relates to U.S. Ser. No. 11/443,668,
filed May 30, 2006, now U.S. Pat. No. 7,539,882, which relates to
U.S. Provisional Application No. 60/685,976, filed May 30, 2005.
U.S. Ser. No. 14/169,464, filed Jan. 31, 2014, also relates to U.S.
Ser. No. 14/090,099, filed Nov. 26, 2013. Each of the mentioned
applications are hereby incorporated by reference in their
entirety.
FIELD OF THE INVENTION
The embodiments disclosed herein relate to power management in
self-powered electronic devices, and, in particular, to predictive
power management in self-powered electronic devices that include
power saving features using activity costs.
BACKGROUND OF THE INVENTION
Private and public networks, such as the Internet, continue to grow
at an exponential rate. These rapidly expanding networks consume a
tremendous amount of power, which is typically supplied from
traditional electrical power grids. As such, most networked
computers are either tethered to electrical wall outlets or require
periodic charging at electrical wall outlets. Many networked
computers, however, have benefitted tremendously by being
self-powered. For example, emergency communication systems,
computing devices in rural areas where access to traditional power
sources is limited, etc.
However, current self-powered devices, such as devices that obtain
energy from solar power, typically waste much of their power
communicating with other networked computing devices. As such, it
would be highly desirable to provide a self-powered device (SPD)
that consumes less power when communicating with other devices in a
network.
Wireless sensor networks have been proposed for a wide range of
monitoring applications in various industries, such as health care,
energy, transportation, infrastructure, agriculture, security, the
environment and many other fields. Billions of active sensors have
been installed that can wirelessly connect to networks and capture
data. Such an installed base of sensors creates issues such as
powering such sensors as well as transporting and storing the data
received from the sensors.
Various conventional approaches have been developed in response to
particular constraints. Microelectronics companies are concerned
with creating low power devices with energy harvesting technologies
and improved batteries. Communications companies are continuing to
build networks with more radio towers, improved data compression
and adherence to interference regulations. Cloud computing
companies are developing approaches for storing more data with data
structures suited for the expected exponential growth in capturing
data. Such conventional approaches are unlikely to be sufficient
with the grand vision of the "Internet of Things". The
communications bandwidth (and data storage on the Internet to
implement such an approach) could include hundreds of times as many
sensors as conventional smartphones, potentially consuming the
highest amount of energy of the Internet.
Most conventional approaches are based on autonomous sensor nodes
that are always on and capture and transmit large amounts of sensor
data to a network that always needs to be listening and recording
the transmissions. The network then needs to relay back an
acknowledgment to the sensor that the data was accurately received.
If scaled to a trillion sensors, a huge number of devices would
need to be implemented with batteries that would be unaffordable
(or at least inconvenient) to change. A huge amount of traffic on
already congested networks would further increase the power needs
of each device, since higher antenna output would likely be needed
to overcome interference and noise.
Different approaches are needed. The motivation of a better
approach stems from a more fundamental question--Why do we need all
of this data anyway? The data and related power needs of capturing,
transmitting and recording data are highly context and application
dependent. Most conventional sensors blindly capture and transmit
data. Such sensors assume the application needs or knows what to do
with that data. Some data inevitably are more important than
others. The importance of data can vary with time and with the
knowledge embedded in the application.
It would be desirable to implement a predictive power management in
a wireless sensor network.
SUMMARY OF THE INVENTION
The present invention concerns a system comprising a plurality of
self-powered devices and at least one remote device. The plurality
of self-powered devices may be configured to (i) perform one or
more tasks and (ii) select one of a plurality of modes of
operation. The remote device may be configured to (a) determine
scheduling data for one or more activities based on a resource
capacity of the self-powered devices and (b) communicate with the
self-powered devices. The activities may comprise one or more of
the tasks. The self-powered devices may determine a computational
cost of performing the tasks.
BRIEF DESCRIPTION OF THE DRAWINGS
For a better understanding of the disclosure herein, reference
should be made to the following detailed description taken in
conjunction with the accompanying drawings, in which:
FIG. 1 is a block diagram illustrating an embodiment of a system
including one or more self-powered devices;
FIG. 2 is a block diagram illustrating a context of the
invention;
FIG. 3 is a block diagram showing an implementation with multiple
sensor nodes;
FIG. 4 is a block diagram illustrating an embodiment of a
self-powered device;
FIG. 5 is a more detailed diagram of one of the sensor nodes;
FIG. 6 is a block diagram illustrating an embodiment of a
photovoltaic collector of FIG. 4;
FIG. 7 is a block diagram illustrating an embodiment of a converter
and storage system of FIG. 4;
FIG. 8 is a block diagram illustrating an embodiment of the current
flow from the collector of FIG. 4;
FIG. 9 illustrates an embodiment of a software protocol stack for a
network interface of FIG. 4;
FIG. 10 illustrates periodic enabling of a transceiver of FIG.
4;
FIG. 11 illustrates coordinated communication;
FIG. 12 is a flow diagram illustrating a low power activation
signal monitoring mode cycles;
FIG. 13 is a flow diagram showing a power-aware data collection
mode cycle;
FIG. 14 is a flow diagram showing a power-aware communications
mode;
FIG. 15 is a context diagram showing wireless signal
configurations;
FIG. 16A is a block diagram illustrating a computer network having
an MS-Exchange server;
FIG. 16B is a block diagram illustrating a computer network having
a web DAV/iCal server;
FIG. 16C is a block diagram illustrating a computer network having
an MS-Exchange server;
FIG. 16D is a block diagram illustrating another computer
network;
FIG. 17 is a flow chart of a method for managing power
consumption;
FIG. 18 is a flow diagram illustrating a method for categorizing
resource availability;
FIG. 19 is a flow diagram illustrating a method for determining
scheduling data;
FIG. 20 is a flow diagram illustrating a method for operating
self-powered devices and/or associated resources based on
scheduling data;
FIG. 21 is a flow diagram illustrating a method for updating
scheduling data for self-powered devices from a remote device;
FIG. 22 is a flow diagram illustrating a method for checking a
power budget of self-powered devices;
FIG. 23 is a flow diagram illustrating a method for determining
task cost for an activity;
FIG. 24 is a flow diagram illustrating a method for updating an
activity cost based on the self-powered device;
FIG. 25 is a flow diagram illustrating a method for coalescing data
for communication;
FIG. 26 is a flow diagram illustrating a method for reserving
capacity in a self-powered device for critical tasks; and
FIG. 27 is a flow diagram illustrating a method for implementing
device coalescence.
Like reference numerals refer to the same or similar components
throughout the several views of the drawings.
DETAILED DESCRIPTION OF EMBODIMENTS
Embodiments of self-powered systems, devices, and methods are
described herein. The self-powered system may include a computing
device that selectively communicates with a self-powered device.
The self-powered device may include: a circuit to power the
self-powered device; memory; a processor; and a program mechanism
that is stored in the memory for execution by the processor. The
program mechanism may include instructions for selecting one of a
plurality of modes of operation, including a first mode of
operation in which the self-powered device consumes less than a
pre-determined amount of power and a second mode of operation in
which self-powered device consumes more than the pre-determined
amount of power. (See FIG. 17).
Embodiments of the invention provide a context-aware and/or
power-aware wireless sensor network system that enables application
needs to determine which data is captured, transmitted and/or
stored at the edge of the wireless sensor network. Such an
implementation may result in lower power usage, lower bandwidth
usage, less interference, and/or less storage and/or processing
capacity in the cloud. The determination of which data is important
may change and may be determined based on context and/or knowledge.
The sensor system is highly adaptable. A software application can
reprogram the sensor nodes with new instructions as data needs
change and/or grow more complex. Additionally, the power management
of the sensor nodes is application-driven to prioritize limited
power resources according to the application needs. Power resources
may be calculated based on available energy, a power budget,
predicted future power needs and/or energy harvesting
opportunities.
The program mechanism may further include instructions for
communicating with the computing device when the self-powered
device is in the second mode of operation. The self-powered device
may conserve power by alternating between the first mode of
operation and the second mode of operation such that the
self-powered device is in the second mode of operation during
pre-determined time intervals. A duty-cycle of the second mode of
operation relative to the first mode of operation may be based on a
power budget for the self-powered device.
The computing device and the self-powered device may be
synchronized such that transmit circuits and receive circuits in
the computing device and the self-powered device are to communicate
signals during one or more of the pre-determined time intervals or
adaptively-determined time intervals. In some embodiments, the
synchronization includes use of a Network Time Protocol. In some
embodiments, the synchronization includes a Wi-Fi periodic beacon
signal.
The computing device may poll the self-powered device during one or
more of the pre-determined or adaptively-determined time
intervals.
The self-powered device may switch to the second mode of operation
prior to one or more of the pre-determined time intervals or
adaptively-determined time intervals, and the self-powered device
may switch to the first mode of operation after one or more of the
pre-determined or adaptively-determined time intervals.
In some embodiments, the self-powered device further includes
instructions for determining a periodicity of the pre-determined
time intervals or adaptively-determined time intervals in
accordance with signals provided by the computing device, and the
duty-cycle is further determined in accordance with the
periodicity.
In some embodiments, the self-powered device further includes
instructions for combining messages that are to be communicated to
the computing device in order to reduce a communication
overhead.
In some embodiments, the system further includes one or more
additional self-powered devices and a coordination element. The
coordination element may assign communication priorities to the
self-powered device and the one or more additional self-powered
devices in accordance with power budgets for these devices.
In some embodiments, the self-powered device further includes
instructions for providing an acknowledgment message to a sending
self-powered device with a higher priority than other types of
messages.
In some embodiments, a circuit converts an external energy source
into signals that power the self-powered device. The external
energy source may include light.
In some embodiments, the circuit includes a component, such as a
photovoltaic cell, a wind energy generator, a thermo-electric
energy generator, a kinetic energy generator, a piezoelectric
device, a magnetic energy generator, and/or a
chemical-to-electricity generator.
In some embodiments, the computing device provides control
information to the self-powered device.
In some embodiments, the self-powered device includes one or more
sensors. The sensor may measure a characteristic, such as
temperature, humidity, pressure, vibration, light, motion, sound,
proximity, flow rate, electrical voltage, and electrical
current.
In some embodiments, the self-powered device may provide
configuration information to one or more output devices in
accordance with the measured characteristic, the internal status of
the self-powered device, any received message, elapsed time, local
and/or remote user interaction, or a local and/or remote automatic
process. The output devices may include a switch, a relay, an
electromechanical actuator, a heater, a fan, a speaker, a solenoid,
a motor, an electric field generator, a magnetic field generator,
and an electro-magnetic field generator.
Reference will now be made in detail to various embodiments,
examples of which are illustrated in the accompanying drawings. In
the following detailed description, numerous specific details are
set forth in order to provide a thorough understanding of the
subject matter presented herein. However, it will be apparent to
one of ordinary skill in the art that the subject matter may be
practiced without these specific details. In other instances,
well-known methods, procedures, components, and circuits have not
been described in detail so as not to unnecessarily obscure aspects
of the embodiments.
The following description describes various self-powered devices
("SPDs"), systems that include SPDs, and methods for operating SPDs
and the systems that they operate in. SPDs are devices that are
configured to operate solely under power sources from external
energy sources, such as light, thermal, kinetic, wind, wave or
other energy. In some embodiments, the SPDs may also be powered by
battery power and/or using a power line at .about.50 or .about.60
Hz.
The SPDs may include a variety of features and/or functionality
(i.e., hardware and/or software) to conserve power. In particular,
the SPDs may use duty-cycle power management in which the SPDs
alternate between low and high-power modes of operation based on a
power budget. The SPDs may also alternate between many different
power-consumption modes. Furthermore, the SPDs may communicate with
computing devices during communication windows. This communication
may be periodic and may utilize pre-determined time windows. For
systems that include multiple SPDs, a communication priority may be
assigned to a given SPD based on its power budget, with a higher
priority accorded to devices that have a lower power budget. The
SPDs may also combine messages in order to reduce communication
overhead and/or give higher priority to providing acknowledgment
messages over other types of messages.
FIG. 1 is a block diagram of a system 100 including one or more
self-powered devices 102. The system 100 may also include other
computing devices, such as desktop or stationary computers 104,
laptop computers 106, personal digital assistants 116, cellular
telephones, or the like. These computing devices may be connected
to one another via one or more routers or switches 110, firewalls
112, wireless access points 114, fixed network connections 108, or
other computer networking equipment, such as computers, cables,
switches, firewalls, routers, bridges, gateways, or the like. These
computing devices are generally located remotely from the SPDs
102.
In some embodiments, the system 100 has a single connection link
between a SPD (such as SPD 102-1) and a remote computing device
(such as computing device 104-1), while in other embodiments there
may be many wired and/or wireless links between one or more of the
SPDs 102 and the computing devices 104, such as Universal Serial
Bus (USB), Firewire, Ethernet, coaxial cable, copper telephone
line, optical fiber, wireless, infra-red, or the like. For example,
SPD 102-2 may couple to the system 100 via a WiFi (Wireless
Fidelity--IEEE 802.11x wireless networking protocol) access point
114. Also, the system 100 may communicate over private or public
networks, such as an intranet and/or the Internet.
Referring to FIG. 2, a block diagram of a system 150 is shown
implementing a context of the present invention. The system is
shown comprising a block 170, a block 172, a block 174, and a block
102. The block 170 may be implemented as a mobile computing device.
The block 172 may be implemented as a network data services device.
The block 174 may be implemented as a network computing services
device. The block 102 may be implemented as a self-powered device,
such as a wireless sensor node. In one example, the SPD 102 may be
implemented as part of the mobile computing device 170. The SPD 102
is shown receiving inputs from a number of blocks 160a-160n. The
blocks 160a-160n may be implemented as sensors. The sensors
160a-160n may also be referenced as sensors A, B, C, etc. The block
160a may represent (or capture) a contextual logic signal. Another
block (e.g., 160b) may represent (or capture) a beacon signal. The
block 160n may represent (or capture) a proximity signal. Various
other signals may be captured by the sensor node 102.
Referring to FIG. 3, a diagram illustrating a plurality of sensor
nodes 102a-102n is shown. The sensor nodes 102a-102n may be
implemented as SPDs. The sensor nodes 102a-102n are shown connected
to a mobile computing device 170b. However, some of the sensor
nodes 102a-102n may be connected to a mobile computing device 170a,
while others may be connected to the mobile computing device 170b,
while still others may be connected to a mobile computing device
170n. The particular number of sensor nodes 102a-102n and wireless
computing devices 170a-170n may be varied to meet the design
criteria of a particular implementation.
Each of the sensor nodes 102a-102n may have a power budget based on
predicted energy input and/or energy output needs. For example, an
energy input need may be a function of past patterns and/or cycles
as well as predictable future opportunities to harvest energy from
the environment (e.g., weather and/or day/night/seasonal patterns
on a photovoltaic device, etc.), at least up to the storage
capacity. The predicted energy expenditures may also be based on
the predicted power consumed given known upcoming calendar and/or
event data as well as other inputs.
A context-aware activation signal may be used to wake up the device
102. The device 102 may begin sensing and/or processing other data,
and/or provide power management that is context-aware. A higher
priority may be placed on processing certain data. Such data may be
data deciding whether to spend more power to transfer the data with
less delay and/or more reliability. For example, logic may be
implemented to boost radio power based on priority and/or distance
from a cellular tower. Such logic may minimize the radio power that
may otherwise reduce interference with other sensors.
One or more of sensors 160a-160n may be implemented internally as a
beacon signal sensor or as a radiation sensor, a blood glucose
sensor, a proximity sensor, chemical/gas sensor, blood pressure
sensor, a location sensor, a pollution sensor, a heart rate sensor,
a vibration sensor, an air flow sensor, a heart rate variability
sensor, an acceleration sensor, a fluid flow sensor, a skin
conductance sensor, a position sensor, a direction sensor, a
rotation sensor, a weight sensor, a displacement sensor, a machine
operations sensor, a fluid retention sensor, a velocity sensor, a
leakage sensor, a respiration sensor, a magnetic field, a
microphone, an ECG, one or more derived quantities, temperature,
image, color, pulse oximetry, physical pressure, odor, drug
delivery, air pressure, voltage, container opening, etc. One or
more of the sensors 160a-160n, or a combination of the sensors
160a-160n may be implemented internally as part of the sensor node
102 (e.g., within a sensor node housing), or alternatively may be
implemented externally (e.g., as a separate sensing device coupled
to the sensor node 102). Additionally, the sensors 160a-160n may
transmit data directly to a network, in which case the sensor node
102 may receive data as a network data source. Such network data
sources may include, for example, environmental or weather data,
location-based data, proximity data, etc.
The various sensors 160a-160n may be configured in one of a number
of categories, such as a logical condition, a fluid/gas level, a
biological process, etc. The logical conditions may be further
configured to receive signals from devices such as a beacon, a
proximity sensor, a location sensor, a vibration sensor, an
acceleration sensor, a position/direction sensor, a displacement
velocity sensor, a magnetic field sensor, a temperature sensor, a
physical pressure sensor, an air pressure sensor, a force/strain
sensor, a moisture/humidity sensor, etc. The fluid/gas level
category of sensors may be implemented to receive signals from
devices such as a radiation sensor, a chemical/gas sensor, a
pollution sensor, an air flow sensor, a fluid flow sensor, a
rotation sensor, a machine operations sensor, a leakage sensor, a
microphone, an image/color sensor, an odor sensor, a voltage
sensor, an electrical sensor, a current sensor, a gyroscope, etc.
The biological process category of sensors may be configured to
receive signals from devices such as a blood glucose sensor, a
blood pressure sensor, a heart rate sensor, a heart rate
variability sensor, a skin conductance sensor, a weight measuring
device, a fluid retention sensor, a respiration device, and ECG, a
pulse oximetry device, a drug delivery device, a container opening
device, a medication dispensing device, a microbial device,
etc.
FIG. 4 is a block diagram of an embodiment of a self-powered
devices 102 shown in FIG. 1. SPD 200 contains a plurality of
components, such as at least one central processing unit (CPU) 202,
memory 204, an optional wireless transceiver 206 coupled to a
wireless antenna 208, an energy collector 210, an energy converter
212 with an optional energy storage device 214, optional user input
and/or output devices 216, an optional network interface 218, one
or more optional sensors and/or output devices like actuators or
one or more sensor/actuator interfaces 220 configured to be coupled
to the one or more sensors, and at least one bus 222 that connects
the aforementioned components. Different embodiments may include
some or all of these components. Also in some embodiments, these
components are, at least partially, housed within a housing, where
the housing may be configured to withstand direct exposure to the
elements, such as sun and rain.
The CPU 202 may include programmable or non-programmable circuits,
such as ASICs or microcontrollers. This circuitry typically
includes non-volatile memory to retain programmed memory 204
functionality, event logs, and/or data, even after a period of
power that is insufficient for continued operation. In some
embodiments, the data contained in memory 204 may be updated while
the SPD is deployed, in response to either local and/or remote
instigation, allowing new and/or different capabilities to be added
to the functionality.
Using the memory 204, the CPU 202 operates and manages the
remainder of the components in the SPD 200. In some embodiments,
some SPDs may also operate and manage other SPDs in the system 100
(FIG. 1).
The wireless transceiver 206 includes a transmitter and receiver
configured for transmitting and receiving radio signals via the
antenna 208.
The energy collector 210 is any suitable mechanism for collecting
energy from the surrounding environment in which the SPD is
located. For example, the energy collector 210 may include
photovoltaic cells that generate electrical energy from light, wind
energy generator(s), thermo-electric energy generator(s), kinetic
(motion) energy generator(s) (such as piezoelectric devices),
magnetic/inductive energy generator(s), chemical-to-electricity
generator(s), etc. Furthermore, the energy collector 210 may
collect electromagnetic energy from, for example, a system of
antenna(s) and rectifying device(s), or other suitable means. In
addition, in some embodiments, electrical energy may be collected
via signals received through the network interface 218. In other
embodiments, energy may be collected from more than one of these
sources at the same time, or at different times. Also in some
embodiments, energy collection may be supplemented with more
traditional power sources, such as replaceable batteries, AC (line)
power, an external DC power supply, Power-over-Ethernet, or the
like.
In embodiments where the collector 210 generates power from light,
the collector may include one or more photovoltaic (PV) cells (also
known as "solar cells"). These cells absorb incident radiation
(light) and convert that radiation to electrical energy. In
addition to the PV cell(s) themselves, the collector may contain
any necessary interconnections to form a series and/or parallel
array of cells. The properties of the electrical energy generated
by a PV cell for a particular intensity of incident light are
typically characterized by an open-circuit voltage V.sub.OC, a
short-circuit current I.sub.SC, and a maximum power delivered
P.sub.MAX, which occurs at a particular output voltage V.sub.MAX
(where V.sub.MAX.ltoreq.V.sub.OC) and output current I.sub.MAX
(where I.sub.MAX.ltoreq.I.sub.SC). (For many PV cells, I.sub.SC and
I.sub.MAX are roughly proportional to the intensity of the incident
light across several orders of intensity magnitude, whereas
V.sub.SC and V.sub.MAX are comparatively constant, perhaps changing
by a factor of only 2 or 3 across the same large variation of
incident lighting conditions).
The PV cells may be based on a number of different technologies,
including monocrystalline silicon, multicrystalline silicon,
amorphous silicon, thin-film, or other photovoltaic technologies.
If the collector 210 uses more than one cell, the cells may be
connected together in series to provide a higher voltage than a
single cell, in parallel to provide a higher current, or in a
hybrid configuration to provide higher current and voltage than a
single cell. The cells may be physically arranged at the best
possible location to receive incident light, such as above the SPD.
In some embodiments, the cells form an integral part of the SPD
housing so as to not detract from the appearance of the device.
Also in some embodiments, energy collection procedures 236
(described further below) are configured to notify the user of the
power output from the collector 210, such that the user may
determine the best physical location of the cells for optimal
energy collection. For example, the SPD may display an indication
or emit an audible tone that varies and/or is representational of
the instantaneous collected power. Since the SPD may have the
ability to store excess collected energy, the ability to maximize
the collected energy over some extended time may be particularly
valuable for some SPD applications.
In some embodiments, the SPD 200 is intended to be used over a wide
range of lighting conditions, varying from direct sunlight to
low-level indoor fluorescent lighting. In these embodiments, the
SPD 200 may efficiently collect and convert the incident light
energy to usable electricity. Selection of the appropriate type and
size of PV cell(s) to use for a particular embodiment of the SPD
200 is typically determined by the corresponding power required,
expected radiant light conditions, and cost constraints for that
embodiment of the SPD 200.
The energy converter 212 is electrically coupled to the collector
210 and the optional energy storage device 214. The converter 212
converts the energy collected by the collector 210 into a form that
is usable by the SPD 200. Typically, this conversion is from one
voltage to another, for example from the maximum power voltage
(Vmax) of the PV cell(s) to the appropriate supply voltage(s) of
the rest of the subsystems in the SPD 200. The storage device 214,
which may more generally include an energy storage device such as a
battery or capacitor, stores the converted energy for later use by
the SPD 200. When excess energy is collected, this excess energy is
stored until the battery 214 is fully charged. When there is
insufficient energy being collected, the battery 214 provides
additional energy until it has discharged or reached a minimum
useable level. If the energy from the collector 210 is otherwise
deemed adequate for use by the SPD 200, then the battery 214 of the
collector may be minimized or eliminated.
For some embodiments, the SPD 200 may need a useful life of many
years, and the converter 212 may need to operate over this lifetime
without requiring replacement of any internal energy storage
device(s), such as the battery 214. This lifetime may need to be
maintained over the expected range of operating temperature and
humidity, and also after many charge/discharge cycles, in some
cases many of these cycles occur every day.
The network interface 218 and/or wireless transceiver 206 may
include hardware as well as associated firmware and/or software
components. The network interface 218 provides communications
connectivity to the remainder of the system 100 (FIG. 1) via wired
connections, while the wireless transceiver 206 and antenna 208
provide communications connectivity to the remainder of the system
100 (FIG. 1) via wireless connections. Some embodiments may include
both wired and wireless connectivity, while other embodiments may
include one or the other. Furthermore, some embodiments may include
more than one network interface 218 and/or wireless transceiver
206. In still other embodiments, no network interface 218 or
wireless transceiver 206 is provided, as the SPD 200 does not
require connectivity to the system 100 (FIG. 1), such as for
stand-alone, sense-and-display or sense-and-control
applications.
The network interface 218 may be of any suitable type, such as a
parallel interface, a serial interface, or the like. These
interfaces may be based on a proprietary format or on
industry-standard protocols, such as SCSI, printer-style Parallel
Port, ESDI, GPIB, PCI, RS-232, Ethernet, I2C, USB, FireWire, Token
Ring, DS1, or DS3. For purposes of this discussion, the possible
wired connectivity also includes communication through optical
fibers. These optical-fiber-based communications protocols may
similarly be serial or parallel, and may be based on industry
standards (e.g.--Fiber Channel, OC-3, OC-12, etc.) or may be
proprietary in nature.
Possible wireless connectivity for the wireless transceiver 206
include radio-based protocols, light-based protocols,
magnetic-induction-based protocols, or other methods of
communicating that do not require some sort of physical connection.
Radio-based interfaces may be based on industry standards, such as
Wi-Fi (IEEE 802.11-based protocols), BlueTooth, RFID, and other
interfaces known to those skilled in the art, or may be proprietary
in nature. Light-based protocols may be industry standard
(e.g.--IrDA, or other industry-standard protocols) or may be
proprietary in nature.
The user input and/or output devices 216 may include input devices
for controlling the SPD 200, such as one or more buttons or
switches, a keyboard, a touch screen, a proximity sensor, a
microphone, and other input functions known to those skilled in the
art, or some combination of these. The user input and/or output
devices 216 may also include output devices, such as one or more
indicators (such as light-emitting diodes, or passive reflective
indicators), multi-digit numeric or alpha-numeric displays, an
array of pixels (such as bit-mapped LCD screen or another pixeled
array of indicators), speakers, or the like. Inputs and outputs may
be initialized and/or updated, through an automatic process and/or
through user interaction, and coordinated remotely and/or locally
at the installed location.
A number of techniques can be used to conserve the power of the
user input/output devices, including operating the output devices
at a lower duty cycle or turning the output devices off when
available power is low. Depending on the underlying technology,
displays may also be dimmed, or operated at reduced contrast or
refresh rate, with different encoding/decoding parameters, and
other power-conservation techniques known to those skilled in the
art. In addition, various forms of input and output functions may
be combined, such as the use of proximity sensors as a "user input"
so that the display may activate automatically when sensing a
user's presence.
The sensory I/O subsystem 220, if present, may include input and/or
output functionality. For example, the sensors may include input
device(s), such as temperature, humidity, pressure, vibration,
light, motion, sound, proximity, flow rate, electrical voltage,
and/or electrical current sensors; and/or output device(s), such as
relays, electromechanical actuators, heaters, fans, speakers,
solenoids, motors, electrical and/or magnetic field generators.
The memory 204 may comprise Random Access Memory (RAM) and/or Read
Only Memory (ROM). The memory 204 may include an operating system
226 that has instructions for communicating, processing, accessing,
storing, or searching data. Examples of suitable operating systems
include embedded LINUX or a proprietary or scaled-down operating
system. In addition, memory 204 may include communication
procedures 228, wireless procedures 230, power management
procedures 232, sensor procedures 234, the energy collection
procedures 236, input/output procedures 238, and other
applications, such as applications 240 and 242.
The communication procedures 228 are used for communicating with
the remainder of the devices in the system 100 (FIG. 1). The
communication procedures 228 may also control power management of
the wireless transceiver 206 and/or network interface 218, protocol
implementation and participation (such as with 802.11b, etc.),
persistent network interface or wireless operating parameter
management, network interface statistics management,
bridging/routing functions, encryption and security services,
etc.
The wireless procedures 230 may work in conjunction with the
communication procedures 228 to facilitate wireless communication
between other devices in the system 100 (FIG. 1) and the SPD 200.
In some embodiments, the SPD 200 may also include a wireless
network adapter, for example, as at least part of a printed circuit
board that controls communication at the data link level (OSI
layers 1 and 2) between an external computing device and the SPD
200.
For some embodiments of the SPD, conserving power is of utmost
importance. In these embodiments, numerous techniques may be used
to reduce power consumption. One or more of these techniques may be
implemented using hardware and/or software, such as the power
management procedures 232. For example, power consumption may be
reduced by maximizing gain of the passive input components;
coalescing messages; explicit attention to latencies; access point
placement; duty cycling and/or sleep modes; priority ranking based
on power levels; static and/or dynamic tokenization of commonly
exchanged information blocks; static and/or dynamic distillation of
repeated information blocks; periodic (versus continuous)
transceiver enable; peer-forwarding based on power disparities;
gateway routing based on power disparities; and other
power-conservation techniques known to those skilled in the
art.
Thus, in some embodiments, the power management procedures 232 are
used to reduce power consumption by the SPD 200 by (i) initiating a
series of prioritized shutdowns of partial or entire subsystems
when the available power from the collector 210 and/or battery 214
becomes insufficient to keep the SPD 200 fully powered, and/or (ii)
ensuring partial or entire subsystems are powered back on when
sufficient power again becomes available to support additional
operation usually according to the priority of each function in the
subsystem.
The sensor procedures 234 are used to measure variables from the
sensors and/or communicate with output devices. For example, the
procedures may be monitoring the sensors for ambient and/or
external stimuli, collecting and formatting the information, and
then initiating any appropriate action based on that input
information. For the output function(s), these procedures initiate,
modify, and/or terminate the operation of the output device(s). In
conjunction with the power management procedures 232, the sensor
procedures 234 also manage the power consumption and operating
characteristics of the sensors and/or the output devices 220.
The energy collection procedures 236 are used to control and manage
energy collection and storage, such as by (i) monitoring the
voltage and current produced by the collector 210 and, based on the
known properties of the collector 210, determining the intensity of
the input energy (light, heat, motion, etc.) to the collector 210
as well as the overall energy being made available to the SPD 200,
(ii) ensuring that the current drawn from the collector 210 is
adjusted to maximize the power extracted from the collector 210,
(iii) ensuring that the storage device 214 (if present) is properly
charged, and/or (iv) monitoring and controlling the available
energy stored in the battery 214, etc.
The user input/output procedures 238 control and manage the user
input/output devices 216, such as by collecting and formatting
(thresholding, debouncing, etc) the user input, and initiating any
appropriate action based on that input. For the output function(s),
these procedures are responsible for any necessary
formatting/construction of the information to be displayed, as well
as updates of that displayed information. In addition to these
primary functions, these procedures in conjunction with the power
management procedures 232 manage the power consumption of the user
input/output devices 216.
Referring to FIG. 5, a more detailed diagram of one of the sensor
nodes 102a is shown as a block (or circuit) 200a. Each of the
self-powered sensor nodes 102a-102n may include a processor, an RF
transceiver, an energy storage unit, one or more sensors and a
memory. The memory may store a set of instructions executable by
the processor to evaluate and selectively store sensed data in the
memory or transmit sensed data based on sensor data values and the
stored instructions. The remote computing device 170 may be located
separately from the sensor node 102a. The device 170 may be
selectively coupled to one of the sensor nodes 102a-102n to collect
stored data and/or to transmit new instructions to the sensor
node.
The sensor node 200a is shown comprising a block 202a, a block
204a, a block 206a, a block 208a, a block 210a, a block 214a, a
block 250a, a block 252a, and a block 254a. The block 202a may be
implemented as a CPU (or processor) circuit. The block 204a may be
implemented as a memory. Details of the memory 204a are shown by
the blocks 270a-270n. The block 206a may be implemented as a radio.
The block 208a may be implemented as an antenna. The antenna 208a
and the radio 206a may form a transceiver. The block 210a may be
implemented as an energy capture unit (ECU). The block 214a may be
implemented as an energy storage unit (ESU). The block 250a may be
implemented as a sensor I/O. The block 252a may be implemented as a
user I/O circuit. The block 254a may be implemented as a clock
generation circuit.
The output portion of the user I/O 252a may be configured to preset
one or more signals to a speaker, a display, an LED, a vibrator, an
actuator, an electrical field, a magnetic field, etc. The signals
presented to the I/O 252a may be an electrical signal (such as a DC
signal), a data signal, etc. The input portion of the user I/O 252a
may be configured to receive signals from a microphone, a camera, a
light sensor, a motion sensor, a position sensor, a button, a touch
screen, a proximity sensor, etc. The signals presented to the I/O
252a may be an electrical signal, such as a DC signal, a data
signal, etc.
In one example, the sensor 102 may implement a learning process
and/or predictive modeling on the CPU 202a (to be described in more
detail in connection with FIGS. 12-14). New rules may be stored in
the memory 204a. In another example, the machine learning
process(es) may be run on the mobile computing device 170 and/or
the network computing services block 174. The processing power and
energy to implement the learning process(es) may use more
aggregated data and/or processing power than the sensor 102 can
process efficiently while running on the energy storage unit 214a.
By offloading the processing, the overall energy used by the
sensors 102a-102n may be reduced.
FIG. 6 is a block diagram of a photovoltaic collector 300 of a
self-powered device, such as the SPD 200 (FIG. 4). As described
above, in some embodiments the collector 300 includes one or more
PV cells 301 connected together to supply power to the converter
212 (FIG. 4). When illuminated, the interconnected PV cells 301
together produce a current I.sub.PV at a voltage V.sub.PV. In other
embodiments, other forms of energy collection and generation of
electricity from external energy sources may be used. Whether the
generator(s) derive their input power from light, heat, motion,
chemical energy, vibration, pressure, network interface
communications, other electromagnetic radiation, or some other form
of energy, or some combination of these external sources, the
electricity produced is characterized by an output voltage at some
output current.
An optimal operating point typically exists where the output
voltage at a particular output current yields a maximum power from
the available input energy and operation at this optimal operating
point may be achieved using the methods described below.
Furthermore, more than one of these PV cells 301 may be combined as
shown using PV cells of the same or different types. Furthermore,
in some embodiments, switching mechanisms are provided so that the
presence of one set of PV cells 301 do not adversely affect the
operation of another set of PV cells 301, such as when one set is
operating, and another set is not operating.
FIG. 7 is a block diagram illustrating an embodiment of a converter
400, such as the converter 212 (FIG. 4). In some embodiments, the
converter 400 includes an input DC-to-DC converter 402, a storage
device 404, and an output DC-to-DC converter 406. The input
DC-to-DC converter 402 transforms the current coming from the
collector at V.sub.PV to the voltage V.sub.STORE of the storage
device 404. The output DC-to-DC converter 406 transforms the
current flowing from the storage device 404 to the voltage(s) used
by the other subsystems in the SPD, known as V.sub.OUT1,
V.sub.OUT2, etc. These output voltages may be the same or different
depending on the requirements of the SPD. If present, the storage
device 404 includes one or more physical rechargeable storage
devices interconnected in series and/or parallel that meet the
requirements of that embodiment. One should note that in some
embodiments, a storage device may not be required. In these
embodiments, all that is required is to convert the input voltage
derived from the collector 210 (FIG. 4) to the output voltage
required by the rest of the subsystems in the SPD. Additionally,
other embodiments do not require the input DC-to-DC converter 406
if the optimal voltage supplied by the collector 210 (FIG. 4) is
within a predefined acceptable range. Likewise, some embodiments do
not require the output DC-to-DC converter 406 if the rest of the
subsystems in the SPD are able to operate using the output voltage
from the storage device 404.
FIG. 8 is a block diagram illustrating an embodiment of the current
flow 500 from the collector 210 (FIG. 4). In order to maximize the
power taken from the collector 210 (FIG. 4), the converter 212
(FIG. 4) attempts to maintain the current flowing from the PV cells
301 (FIG. 6) at about I.sub.MAX, which yields a voltage of
approximately V.sub.MAX. Since the optimal voltage V.sub.MAX of the
collector 210 (FIG. 4) may be different from the voltage
V.sub.STORE of the storage device 404 (FIG. 7), some embodiments
may make use of a DC-to-DC converter (as described above) to make
the voltage transformation, preferably with minimal power loss.
Suitable DC-to-DC converters operate in a switching mode, which is
to say that they periodically draw some current to "charge" an
inductor, then "discharge" that current (at a different voltage)
into the load. These two operations may then be repeated when more
power needs to be transferred.
For DC-to-DC converters, the time that the current is being drawn
into the converter to charge the inductor may be referred to as
T.sub.ON, and the time that the DC-to-DC converter is transferring
the stored charge to its load (or is otherwise waiting for the next
cycle to begin) may be referred to as T.sub.OFF. The full cycle
time of the converter may be expressed as T.sub.ON+T.sub.OFF.
Similarly, the current drawn into the DC-to-DC converter when it is
on may be referred to as I.sub.DCON, and the current drawn into it
when it is discharging its inductor to the load (or it is otherwise
waiting for the next cycle to begin) may be referred to as
I.sub.DCOFF. In order to maintain the maximum power output from the
collector, the time average of these two converter input currents
may be kept to about I.sub.MAX.
For most switching DC-to-DC converters, I.sub.DCOFF is
approximately equal to 0, and assuming that the converter is on for
only a small fraction of its full cycle time (T.sub.ON+T.sub.OFF),
the I.sub.DCON usually needs to be greater than I.sub.MAX.
Unfortunately, for a DC-to-DC converter operating in these
conditions connected directly to the collector 210 (FIG. 4), the
current drawn alternates between I.sub.DCON during T.sub.ON and
approximately 0 during T.sub.OFF, neither of which are the optimal
current I.sub.MAX. Therefore, to maintain the optimal current drawn
from the collector at about I.sub.MAX, a capacitor 510 or other
charge-storage device may be connected across the output of the
collector, as shown. With the addition of the capacitor 510, when
the DC-to-DC converter is on, the net current into the capacitor is
I.sub.PV-I.sub.DCON. Since I.sub.DCON is typically greater in
magnitude than I.sub.PV, the capacitor 510 is therefore discharging
during T.sub.ON. During T.sub.OFF, typically I.sub.DCOFF.apprxeq.0,
so the net current into the capacitor 510 is
I.sub.PV-I.sub.DCOFF.apprxeq.I.sub.PV, and the capacitor 510 is
charging. The average value of the current into the capacitor 510
may be expressed as
##EQU00001## which may be set to be approximately I.sub.MAX for
efficient operation of the collector 210 (FIG. 4). For large
capacitances, the drop in the voltage of the capacitor 510 during
T.sub.ON may be very small, and the rise in the voltage during
T.sub.OFF may likewise be very small.
Since the average current drawn from the collector 210 (FIG. 4),
I.sub.PV, is now maintained at approximately I.sub.MAX, the average
voltage on the capacitor 510 (and the collector 210 in FIG. 4) is,
therefore, approximately V.sub.MAX. The value of the capacitor 510
is chosen such that the variation in voltage is small enough to be
acceptably close to V.sub.MAX.
As previously stated, the maximum power of one of the PV cells 301
(FIG. 6) typically occurs at a certain current I.sub.MAX, and at a
certain voltage V.sub.MAX. However, the actual values of this
operating point may change with incident light intensity, operating
temperature, and may also vary from device to device due to
processing and manufacturing differences. This may make it
difficult to predetermine the ideal operating point values for any
particular device at any particular light intensity and
temperature. In some embodiments, the ideal operating point for the
PV cell(s) 301 (FIG. 6) may be determined automatically and,
furthermore, may be adjusted automatically as operating conditions
change, using the following method.
To determine the power being delivered by the converter 212 (FIG.
4), the energy collection procedures 236 (FIG. 4) periodically
measure the current I.sub.PV from the collector 210 (FIG. 4), and
the voltage V.sub.PV on the collector 210 (FIG. 4), and multiplies
these two values together. One way to increase the average current
drawn by the input DC-to-DC converter 402, usually resulting in a
reduced V.sub.PV, is by increasing the on time of the DC-to-DC
converter, T.sub.ON. Likewise, by reducing T.sub.ON, the average
current drawn from the collector 210 (FIG. 4) may be decreased,
allowing V.sub.PV to increase. Other means to increase or decrease
the average current drawn by the input DC-to DC converter may be
used.
To maintain the maximum power from the collector 210 (FIG. 4) under
varying operating conditions, the energy collection procedures 236
(FIG. 4) (i) measure the power from the collector 210 (FIG. 4),
(ii) increase I.sub.PV by a small amount, such as 0.1% to 1% of the
full value, (iii) measure the power from the collector 210 (FIG.
4), (iv) if power has increased, go to step (ii) above, otherwise
continue, (v) decrease I.sub.PV by a small amount, such as 0.1% to
1% of the full value, (vi) measure the power from the collector 210
(FIG. 4), (vii) if power has increased, go to step (v) above,
otherwise go to step (ii) above. Using this process, the controller
may continuously "hunt" for the maximum power delivered from the
collector 210 (FIG. 4). As the operating conditions vary, this
procedure may allow those changes to be tracked by the SPD,
allowing the energy collection procedures to continuously adjust
the operating point of the collector 210 (FIG. 4) in order to
achieve maximum power delivery.
To prevent operating at only either the maximum current I.sub.SC or
the maximum voltage V.sub.OC, which may happen under very low
lighting and/or other extreme conditions, when it may be difficult
to determine how the power from the collector 210 (FIG. 4) has
actually changed, the procedure may be modified to switch the
direction of the "hunt" if the change in I.sub.PV value has moved
in the same direction for the last 10 or 20 (or other appropriate
number of) steps or iterations. This means that under extreme
operating conditions, the operating point of the collector 210
(FIG. 4) may be scanned across a wide range of possible operating
points to increase the likelihood that the maximum power delivery
point is found. In some embodiments, to ensure that the storage
device 404 (FIG. 7) in the converter 212 (FIG. 4) is maintained
within its safe operating parameters, its voltage and/or current
may be continuously monitored by the controller.
When the maximum capacity of the battery 214 (FIG. 4) is reached,
the controller typically turns off the input DC-to-DC converter.
This may be accomplished by setting the on time T.sub.ON of the
DC-to-DC converter to approximately zero, or otherwise disabling
its operation.
When the energy held in the battery 214 (FIG. 4) gets low, and/or
is not otherwise being sufficiently recharged from the collector
210 (FIG. 4), the energy collection procedures 236 (FIG. 4) may
reduce the power consumption of the other subsystems in the SPD. In
some embodiments, the energy collection procedures 236 (FIG. 4) may
compare the energy in the battery 214 (FIG. 4) against one or more
pre-determined low-energy thresholds, and once the threshold(s) is
crossed, the energy collection procedures 236 (FIG. 4) may take one
or more action(s) to reduce the energy demand from the battery 214
(FIG. 4), if appropriate.
When the remaining energy in the battery 214 (FIG. 4) drops to the
lowest threshold, below which proper operation of the SPD may no
longer be guaranteed, in some embodiments, the energy collection
procedures 236 (FIG. 4) and/or the power management procedures 232
(FIG. 4) initiate a series of controlled shutdowns of any
subsystems that remain on, including the majority of the SPD
itself. Once this is accomplished, the SPD may be held in a reset
condition, through the use of an automatic circuit until sufficient
energy is delivered to the battery 214 (FIG. 4) to allow initiation
of SPD startup.
When the energy level of the battery 214 (FIG. 4) transitions from
below the lowest threshold to above that threshold, a subset of the
SPD may be automatically brought out of reset and may begin to turn
on various subsystems of the SPD, typically in a prioritized
sequence, depending upon how much energy is in the battery 214
(FIG. 4), and/or how quickly it is being recharged. As increasing
energy becomes available in the battery 214 (FIG. 4) and/or more
power is available from the converter 212 (FIG. 4), and the various
low-power thresholds are overcome, more and more SPD subsystems may
be brought out of any low-power states until eventually the SPD may
be returned to its full-power mode.
To eliminate unnecessarily repeated or oscillatory transitions
between various low-power states, some embodiments may include an
appropriate amount of hysteresis to be added to each threshold.
This ensures that as a low-power threshold is approached in the
decreasing direction, a slightly lower value than the nominal
threshold value may be required to be achieved before transitioning
the SPD to the next lower-power state. Likewise, when the energy in
the battery 214 (FIG. 4) is increasing and that threshold is
approached, a slightly higher value than the nominal threshold
value may be required to be achieved before transitioning from the
previous low-power state. The amount of hysteresis for each
threshold level is chosen to minimize oscillatory transitions, yet
not make the actual transition points too different from the
nominal threshold value.
In a given embodiment, there may be zero or more network
interfaces, such as the network interface 218 (FIG. 4) of one or
more types. The specific type(s) of network interfaces utilized in
an SPD do not necessarily imply a specific network behavior, and
may include technologies such as, but not limited to, wired
interfaces (RS-232, RS-485, USB, Ethernet, FDDI, Fiber Channel, or
any other wire or cable-based protocol, standard or proprietary)
and wireless interfaces (Bluetooth, Zigbee, IrDA, UWB, Wi-Fi, GPRS,
GSM, CDMA, or any other wireless protocol, standard or
proprietary).
As illustrated in FIG. 9, typically each network interface includes
the following hardware and software layers, as defined by the
classic OSI networking reference model: physical layer 610 (such as
RS-232, 100BaseT, T1 and radio), link layer 612 (such as Ethernet,
FDDI, Wi-Fi and HDLC), network layer 614 (such as IP, IPX and
similar protocols), transport layer 616 (such as TCP, UDP, RTP and
SMB). Higher layers, such as session layer 618, presentation layer
620 and application layer 622, may be viewed as part of the client
software, so are not discussed as part of the network interface
itself.
In many embodiments, one of the key attributes of the network
interface may be its operation on an extremely low power budget.
This is discussed further below.
Numerous issues may affect power usage in an implementation of the
network interface, including range requirements, the amount of
power consumed per unit of time, the amount of power consumed per
bit conveyed, and any needed networking layer overhead from the
associated network interface software stack and its impact on the
power consumed in generating, transmitting, receiving, and
processing that overhead.
In general, the power consumption of a communication subsystem
increases with the amount of data communicated and the distance
that the communications traverse. Similarly, a low-bit-rate
communications system may generally be designed to consume less
power per unit of operating time than a high-bit-rate system, but
may have a higher cost per bit communicated. Finally, communication
subsystems based on newer, specialized, and/or proprietary
communications protocols often consume less power for a given
transmission distance or bit rate than existing, general-purpose,
and/or standardized protocols.
Keeping in mind these general characteristics, the design,
implementation, and usage of each network interface for a
particular embodiment of an SPD may have various constraints
including (but not limited to): The need for wireless vs. wired
operation. The distance required for data transmission. The volume
of data to be transferred for the SPD application. The sensitivity
of the application to data stream latency. The frequency of
communications needed to meet protocol requirements. Any
compatibility requirements with installed network infrastructure.
The average power available to the network interface subsystem. The
tolerable cost and/or size of the subsystem.
Several methods to reduce average power in the network interface
subsystem are described here. Embodiments of the SPD may use one,
many, all, or none of these methods and strategies. Some strategies
outlined are applicable to any network interface technology while
others are applicable only to one or some network interface
technologies.
One class of radio computer network technology is based on the IEEE
802.11 set of standards, commonly referred to as "Wi-Fi". Versions
of this standard support various transmission distances and bit
rates, and existing equipment may support more than one version
simultaneously. In an exemplary embodiment, therefore, the SPD may
be used with one or more IEEE 802.11 "Wi-Fi" Protocols.
Implementation in embodiments that support a compatible version of
Wi-Fi may typically not require installation of an additional
wireless access point(s) in the external computer network. Instead,
an existing, i.e., an already-installed, access point(s) may
typically be used. In some cases, the access point functionality
may already be incorporated into the remote device(s), such as
computer devices 104 (FIG. 1), with which the SPD(s) 102 (FIG. 1)
communicate.
An additional advantage of using a Wi-Fi radio is that the
components needed to implement the radio are often readily
available, physically small, and relatively inexpensive. The Wi-Fi
data rate may also be high enough for most SPD applications.
However, the existing component sets may consume relatively large
amounts of power, even when not transmitting. Thus, in some
embodiments, various schemes to dramatically reduce the power
consumption of the Wi-Fi radio may be employed.
In some embodiments, the SPD includes sleep modes and/or duty-cycle
control. Selectively shutting down or turning off some or all of
circuits in the physical-layer networking subsystems may offer the
ability to reduce power consumption. These features are typically
referred to as "sleep modes". The duty cycle of such subsystems is
the ratio of the time the subsystem spends in the "on" or "awake"
state compared with the total time. Assuming the higher-layer
protocols and/or applications may tolerate it, adjusting the
duty-cycle of the network interface may allow the controller to
reduce power consumption of the subsystem without requiring
additional external hardware and/or reconfiguration of the
device.
For some network interface technologies, a high-rate of power
consumption when active may be more than offset by the low power
consumption from remaining in sleep mode the majority of the time
and powering the subsystem for the minimum time necessary to affect
communication. Thus, for a given amount of data to be transferred,
some implementations of the network interface may have a high
power-per-unit-of-time cost but may still be efficient in the cost
of the power per bit communicated.
This approach accepts a high power draw for short periods of time
because it seeks to optimize the power per bit communicated through
the network interface. It is possible for such an implementation to
consume less average power than an implementation using a
low-bit-rate protocol because the network interface may spend
substantially less time powered on even though the network
interface is drawing more power when it is powered on.
Often, it is not known beforehand when a message must be sent from
an SPD to a remote device such as one of the computer devices 104
(FIG. 1) and vice-versa. In particular, if the network interface in
an embodiment is operated with duty-cycle control (as described
above and further below) then its transceiver spends some of its
time active, and some of its time disabled in a sleep mode. If a
message is to be properly communicated, the transmitting device
must be enabled and the receiving device must likewise be enabled
in order to detect the message and properly receive it.
When there are no power consumption constraints, the transceiver
circuits in each device may be left enabled all of the time. In
some embodiments, however, there may be power consumption
constraints, so it may be desirable to use duty-cycle control as a
power management technique. To ensure that message communication
may occur reliably under these conditions, the duty cycles between
the transmitting and receiving devices need to be aligned such that
the transceivers are enabled in each device at the same time.
By using periodic communication windows, the duty-cycle alignment
may be achieved through synchronization. This assures that the
transmitter in the sending device and the receiver in the receiving
device are enabled at the same time. This is illustrated in FIG.
10, in which (a) transceivers are enabled just prior to the
communication time window, (b) communication occurs, often lasting
just long enough to initiate a session (see FIG. 9) that will span
numerous enable periods, and (c) each SPD shuts down its
transceiver as soon as it can, but no later than after some limited
period of time.
In some embodiments, therefore, periodic transceiver enabling is
used. Different network interface protocols may support such
enabling. For example, a sparsely populated Wi-Fi network may be
used in conjunction with a periodic beacon signal having an
appropriate rate. The periodic beacon signal may be adjustable. The
periodic beacon signal notifies some or all radio NICs (the Wi-Fi
nomenclature for a remote node versus an access point) of its
presence, timestamp, supported rates, capabilities, and an
indication of any traffic waiting to be transmitted (TIM). In
concert with the beacon signal, a Wi-Fi NIC may be kept in sleep
mode for an interval just short of the beacon signal interval, and
then be automatically woken-up in time to receive the beacon
signal. The NIC may then determine if there is any traffic queued
up for it to receive, and then receive the traffic (if any). It may
then send the pending items it has queued for transmission (if any)
and then go to sleep again until the next beacon signal
interval.
In some embodiments, to reduce the power consumed by a Wi-Fi radio
subsystem that uses commercially-available chip sets, yet still
maintain an adequate communications bit-rate, the radio may be kept
in a low-power "off" or "standby" mode as much as possible. In some
embodiments, the standby mode may be used the majority of the time.
One way to do this is to take advantage of Wi-Fi's beacon feature,
described above, whereby the Wi-Fi access point associated with an
SPD periodically polls the SPD (and any other associated remote
Wi-Fi devices the access point is aware of) at regular pre-set
intervals. The beacon signal interval may often be set at once per
100 ms (although a different interval may instead be configured in
the access point). By monitoring the beacon signal activity, the
SPD determines the beacon signal period. Thus, the SPD knows when
the next beacon signal will occur, thus when it must awaken to
receive the next beacon signal.
By placing the SPD's radio in a low-power "off" or "standby" mode
until just before the beacon signal arrives, the SPD's controller
may ensure that the SPD radio is active during the beacon signal
time so as to transfer any data to or from the access point, and
then return the radio to its low-power state as soon as the data
transfer is complete. Assuming the beacon signal interval is not
set to be unreasonably short, this strategy may allow the SPD radio
to spend the large majority of its time in a very low power state
yet still maintain sufficiently low transmission latencies and
adequate data throughput to support typical SPD operation.
Some techniques used to reduce power consumption in the network may
benefit from having an SPD synchronized in some way with one or
more other nodes and/or access points in the network. If all nodes
and/or access points participating in the communications exchange
are synchronized then they will know when to enable their
transceivers for periodic communications.
In one embodiment, synchronization may be achieved using the
Network Time Protocol (NTP) that is typically available in computer
networks that adhere to the Internet Protocol (IP) set of
standards. By utilizing the NTP facilities available in the
network, an SPD and other nodes may accurately determine the time
of day and, from this information, facilitate their periodic
communications.
In another embodiment, synchronization may be achieved by leaving
the SPD's network interface receiver enabled long enough to receive
several periodic transmissions. By observing the periodicity of the
transmissions, the SPD may then enter a sleep state between the
periodic transmissions, gradually increasing the length of the
intervening sleep states until the duty cycle of the network
interface is low enough to achieve the desired power savings but
high enough to ensure sufficient probability of being awake during
the periodic transmissions.
Moreover, standardized communications between computers typically
operate in layers. However, additional layers increase overhead
costs that negatively impact efficiency, as measured in terms of
payload data communicated per unit of energy and/or power consumed.
Coalescing messages is a systems approach to efficiency, and allows
for communications to be optimized by having additional knowledge
about the communicating devices and/or the protocols they are using
to communicate.
In a typical embodiment, messages for various purposes flow between
the SPD and one or more remote device(s). In the event that more
than one such message is going to the same remote device, it may be
possible to examine the pending messages destined for that remote
device and combine the messages into a form of multi-block message
so that layering overheads are minimized.
Combining or coalescing messages may be combined with periodic
transceiver enabling and duty-cycle control to create a natural
collection or "queuing" point in time where opportunities for
combining the messages prior to communication may be found.
Most messages to and from an SPD may be categorized into various
classes. Attention messages include a message from a client node on
the network requesting information or service from a server node.
Information messages include a message to or from an SPD containing
information. This message may be in response to an Attention
Message and/or may contain unsolicited information. Ack/Nak
Messages include a message to or from an SPD indicating that a
previous message is being acknowledged, or negatively
acknowledged.
In an exemplary embodiment of coalescing, an information block
consisting of 1000 bytes in total of data is to be sent from a
remote device to an SPD. Even though the 1000 bytes are to be sent
from the same remote device to the same SPD, because of the way the
information block is typically created using modern software
methods, it is often sent as a series of smaller messages, each
consisting of 50 to 100 bytes of payload data. Each message, if
sent individually over Ethernet, typically may have a wrapper that
includes the UDP header, the IP header, and the 802.3 header, plus
the preamble. In total, this may double the number of bits
transmitted in each message, as compared with just the bits in the
payload data. Therefore, coalescing these messages into a
multi-block message may require a frame to indicate the coalescing
behavior (perhaps 4 bytes) and an additional approximately 2 bytes
per block of data to demarcate each sub-block for later extraction.
This multi-block message may still have the UDP header, the IP
header, the 802.3 header, and the preamble, but only one for each
of these, thus requiring approximately one tenth the overhead
necessary as compared with no coalescing of the information.
Communicating a message from one computer to another generally
involves some form of acknowledgment (positive or negative) that
the message was seen, acted on, or otherwise processed. When an SPD
sends a message to a remote device, the time that the SPD spends
waiting for acknowledgment to return from the remote device may
often be time that the SPD is waiting with its network interface
and/or other subsystems powered on. Minimizing the time waiting for
the return Ack/Nak Message minimizes the overall time that the SPD
subsystems remain powered on, thus reducing the average power
consumption of the SPD.
There are a variety of ways such latency may be reduced or
minimized, including: Arrange for the remote device to acknowledge
a message immediately upon its receipt by the remote device, and
not to wait for the message to be fully processed and/or any
message-requested action to be completed. Arrange for the remote
device to acknowledge a message as a high-priority task in its
internal task-scheduling algorithm. Allow the remote device to
coalesce the Ack/Nak Messages it needs to return to an SPD and then
let the SPD "decoalesce" them upon receipt for distribution to the
SPD's internal sending entities. The coalescing of Ack/Nak messages
by the remote device may be performed separately for each SPD that
is communicating with the remote device. Process the messages where
the sender is awaiting a response preferentially over other message
types. Some, all, or none of the above methods may be used in an
SPD and/or remote device implementation in order to reduce the
power consumption and/or otherwise increase the operating
efficiency of an SPD.
In applications where a remote server is handling information for
multiple requesting SPDs, there may arise cases where some SPDs
have a lower average power budget than others. In such cases, the
scheduling of resources when dealing with multiple SPDs
simultaneously may favor those with lower power budgets to minimize
the impact of latency on their power budgets.
In the case of multiple SPDs requesting communication in an
environment using periodic transceiver enabling, one technique to
optimize power consumption includes a coordinating element (such as
the server handling the requests) giving preferential permission to
communicate to the energy-poor SPD(s) ahead of the energy-rich
SPD(s) during the period of permitted communication. The
energy-poor SPD(s) may then disable their network interfaces(s)
and/or other subsystems sooner, better conserving their
less-abundant power than the energy-rich SPDs. This approach is
illustrated in FIG. 11, in which (d) transceivers are enabled just
prior to the communication time window, (e) numerous SPDs vie for
the ability to communicate so that a coordinating element (such as
a server) utilizes a token scheme to control which SPDs communicate
thereby allowing energy-poor nodes to communicate preferentially in
order to allow them to disable their transceivers sooner and save
power, and (f) each SPD shuts down its transceiver as soon as it
can, but no later than after some predetermined and limited period
of time.
In the case of multiple SPDs requesting service from a remote
device, another technique to optimize power consumption includes
the server preferentially servicing the power-poor SPD(s) ahead of
the energy-rich SPD(s) if requests from more than one SPD are
pending completion at the server. Minimizing the response time to
the energy-poor SPDs may help them conserve their less-abundant
power by allowing them to place their network interface(s) and/or
other subsystems in a sleep mode sooner at the expense of more
energy-rich SPDs.
As a refinement on the ideas of periodic transceiver enable and
duty-cycle power management, pseudo-time slot allocation may be
used. This technique recognizes that multiple SPDs operating in a
network or ecosystem may experience contention in communicating,
especially to a central point such as a Wi-Fi access point and/or a
server. This technique allows each client SPD to attempt to
exercise its periodic communications during the same periodic
interval but each in a time slot that occurs at a different,
pre-assigned phase within the periodic interval. Just prior to the
periodic communications interval, the master/server node may queue
up time slot assignment messages, one destined to each client SPD,
indicating when the SPD may exercise its right to communicate
within the periodic interval. Alternatively, another method of
assigning time slots may be used as long as only one client SPD is
assigned a particular time slot or as long as there is an effective
way to handle time-slot conflict issues.
At the beginning of the periodic interval, all SPDs accept one (or
a small number of) message(s) to allow the time slot assignment
message to be received. Alternatively, this part of the protocol
may be eliminated if the time slots have already been assigned
through some other method. If the assigned time slot for an SPD has
not yet arrived, the SPD may place its network interface and/or
other subsystems in a sleep state (if they are not already in such
a state) and reawaken them just before the time slot arrives. When
the assigned time slot does arrive, the appropriate SPD may then
transfer any pending messages with the master/server node.
The method(s) used to assign the order of the time slots may use
any number of algorithms or procedures, such as pseudo-random,
round-robin, fairness, and other algorithms known to those skilled
in the art, and may also offer the most energy-poor SPD in the
ecosystem the first right to communicate, followed by the
next-most-energy-poor SPD, and so on in an orderly progression. The
scheduling algorithms may take advantage of information such as the
energy available to the SPD, the estimated energy it may need to
communicate, its communications bandwidth and/or latency
requirements, how often it has communicated in the past, and/or
other relevant information.
In a communications system between computers that implement an
application having a predefined function, blocks of information are
typically communicated repeatedly, and the contents of some of the
blocks may often be known beforehand. For a communications
subsystem where every communicated bit represents another reduction
in the available power, replacing large blocks of this type with
reference tags may reduce the number of bits that must be exchanged
to accomplish a given functionality. Receipt of a particular
reference tag may indicate to the receiving device that it may
replace the reference tag with the associated block of
information.
In an exemplary embodiment, an SPD is used to display a calendar. A
calendar view may be represented by a series of visual "planes".
Each plane may represent something either new to a day's calendar
entry or something that does not change from one day to the next.
The unchanging elements may be considered background planes that
may be stored in the SPD memory and referenced by a symbolic name
(a reference tag) that is shorter than the image they represent.
Typically, when displaying a calendar, the elements that must
change regularly from one day's view to the next are few and are
generally the label for each block of allocated time. All other
items may effectively be referred to by the reference tag,
effectively reducing the number of bits that must be transmitted to
display the calendar view.
An enhancement to traditional static distillation (as described
further below) is a process of dynamic distillation of repeated
blocks. In this dynamic process, a server (either an SPD or a
remote device) may associate a block of information represented by
a reference tag "on the fly". After establishing the meaning of the
reference tag with the client (either a remote device or an SPD),
the server thereafter utilizes the tag to communicate the intended
information in a more cost-effective manner than communicating the
information itself.
In an exemplary embodiment, an SPD is used to display a calendar.
Consider a block of time on the calendar with a label describing
that block of time. If the total calendar time was 8 hours and the
label applied to a 30 minute interval, the label describing that
time frame may typically be transmitted to the SPD, in full,
approximately 16 times. Conversely, sending a single message to
define the label image along with its reference tag may allow use
of just the tag in all subsequent communications, thus reducing the
amount of information transmitted (and possibly stored).
As discussed previously, it is often the case that proprietary
and/or shorter range communication protocols may operate on less
power than standard and/or longer-range protocols. In addition, it
is possible for the different SPDs in a network or ecosystem to
have vastly different power budgets due to available energy
conditions, opportunity for wired operations, or other factors. In
such cases, a gateway/peer-forwarding behavior may be implemented
that utilizes SPD power reserves as a link-cost metric. Combined
with routing behaviors among SPDs in an ecosystem, such a routing
algorithm may provide the optimal communications path by allowing
the energy-rich SPD or SPDs to operate as gateway nodes to other
parts of the network.
In an exemplary embodiment, a Wi-Fi transceiver and a
very-low-power radio transceiver are provided as network interfaces
in each of several SPDs in a network. The SPDs communicate with
each other and with a non-SPD remote device. The Wi-Fi network
interface has a power-per-bit-communicated cost of transceiver
operation that is many times what the same bit costs to communicate
over the very-low-power radio link. Furthermore, at least one of
the SPDs may have an abundant energy source. In such a case, the
SPD that has abundant energy may be selected to communicate through
its Wi-Fi network interface to the remote device, forwarding
messages to and from the other SPDs through its very-low-power
radio link. The other SPDs, which may have relatively scarce energy
sources, may therefore communicate through their very-low-power
radio links, and, therefore, reduce the power consumed for
communications.
In addition, some embodiments of the SPD contain software and/or
hardware mechanisms to aid in the placement of an entire SPD and/or
access point. These mechanisms may be utilized in order to maximize
the quality of the wireless networking signal in order to improve
communications bandwidth, to reduce the communications error rate,
and/or to reduce the power consumed for communications. For
example, in some embodiments, the SPD includes a mechanism that
displays an indication or emits an audible tone that varies
proportionally with and/or is representative of the quality of the
communications signals to aid in the placement of the SPD and/or
access point.
In addition to (or instead of) optimizing the placement of SPD
nodes and/or access points, the existing placements may be utilized
along with a mechanism to additionally take advantage of the nodes
and/or access points with better placements. As an extension to the
concept of gateway operations for wireless networks (described
further below), the SPD may dynamically or statically select which
SPD and/or access point is to be used as the gateway for the others
based on a combination of SPD energy reserves and/or communications
signal strength.
In an exemplary embodiment, obstructions may sometimes severely
influence the signal strength seen by a wireless node on a network.
Moving the antenna, emitter and/or detector for an access point or
SPD by a few feet may make dramatic differences in signal strength.
Treatment of a group of SPDs as a system and selecting the SPD(s)
with the best signal strength to operate as the gateways for the
group of SPDs may allow for higher communications quality and/or
lower communications power.
In some embodiments, another method to reduce the power consumption
of an SPD wireless radio subsystem is to turn the radio on briefly
after a period of "off" time, such as once every 10 or 100 seconds.
Assuming the "on" time during this period is kept to 10 or a few
10s of milliseconds (or less), this may result in a sufficiently
low average power such that even low-cost, readily-available
wireless protocol (such as Wi-Fi) chip sets may be used in SPDs
with extremely tight power constraints. The ratio of "on" time to
"off" time and the corresponding ratio of "on" power to "off" power
determines the average power consumption of the SPD's
communications subsystem. Thus, the "off" time may be chosen to
achieve a particular low average power consumption given a minimum
"on" time to complete the wireless protocol steps needed to send
and/or receive a complete message, the "off" time may be chosen to
be appropriately long.
One difficulty with this approach may be in keeping the access
point (or other centralized wireless protocol device) from
concluding that the remote wireless protocol device (in this case,
the SPD) has left the wireless network. If the access point has
concluded that the SPD has left the network and then subsequently
the SPD attempts to reestablish communications with the access
point, this may require the SPD to go through a complete wireless
protocol "rejoin" procedure each time the SPD wanted to have data
wirelessly transferred. By convincing the access point that the SPD
has been subject to a temporary interference with its transmission
signal, the communications may be reestablished without requiring
the more involved, time-consuming, and power-consuming "rejoin"
procedure. Various schemes known to those skilled in the art may be
used to convince the access point that the SPD has not left the
wireless network, even if the SPD has not communicated for many
seconds.
Some embodiments may have a minimum amount of user input/output
(I/O) functionality for setting basic operating parameters. In some
embodiments, even this minimum amount of user I/O functionality may
be unnecessary for normal operation and this I/O may be made
inaccessible after initial deployment of the device.
In some embodiments, one or more SPDs may include a
single-indicator display. A variety of power-saving techniques may
be applied to a single-indicator display. These include (but are
not limited to): Applying intermittent power to the indicator at a
high-enough frequency such that it appears to be on continuously
when viewed by the human eye. This frequency may be in the range of
60 cycles per second or greater. In concert with intermittently
powering the single indicator, adjust the duty cycle of the power
circuit for the indicator such that the indicator is powered for
substantially less than 100% of each cycle. This technique relies
on the human eye's propensity to integrate the light that it
receives in such a way that a rapidly-pulsing indicator may appear
to be on continuously if the frequency of pulsation is high enough,
and also that the eye will tend to overestimate the width of the
pulses such that the indicator will appear brighter than the duty
cycle may otherwise suggest.
Also in some embodiments, one or more SPDs may include a simple
numeric or alpha-numeric display. A variety of power saving
techniques, such as those described above, may be used with these
types of displays.
In some embodiments, one or more SPDs may include a pixeled
display. A variety of power-saving techniques may be applied to
pixeled displays. These include (but are not limited to):
Application of sleep modes based on scheduled "night" periods.
Leaving the display in sleep mode until user proximity is
automatically detected and/or the user otherwise triggers active
operation of the display. Requiring an explicit user trigger for
high-power-drain items such as backlights.
Depending on the application being served by the SPD, numerous
forms of sensors may be utilized in an SPD embodiment. Sensors with
implications for power management or other unique purposes are
described below, but any form of sensor operable within the SPD's
power budget, size restrictions, product cost, and/or other
constraints may be supported.
In some embodiments, one or more SPDs may include a passive
infra-red (PIR) device or some other mechanism to effect proximity
detection of personnel. Their application in an SPD embodiment for
display of information may alter the behavior of the SPD to
minimize power consumption. For example, driving or updating an LCD
display consumes power that may be reduced or eliminated by
suppressing updates to the display and/or shutting down the display
if no one is present to observe the content of the display.
In some embodiments, one or more SPDs may include a photo-detector,
since the presence of light may be an indicator of current or
pending activity. In one embodiment, to display information, the
absence of light may also indicate it may be prudent to enter a
power-down state. The SPD may also inform a central monitoring
point that insufficient light is present in an area that is
intended to be lit for security, safety, aesthetic, or other
reasons.
A variety of methods and strategies may be used to reduce power in
the controller. Some embodiments of the SPD may use one, many, all,
or none of these methods and strategies.
In some embodiments, operating power is reduced during periods of
inactivity. Often times the controller's periodic tasks may be
arranged such that they are completed in a burst of activity
followed by a period of inactivity until the periodic tasks again
need to be performed. As long as these periods of activity and
inactivity are kept relatively short and frequent, it may appear to
the rest of the SPD that the controller is performing these tasks
continuously, albeit at a slower rate.
In some embodiments, the controller may be designed to "sleep"
during the periods of inactivity, and "wake up" in response to some
stimulus, change of state, and/or elapsed time. This may be useful
when the power consumption of the controller during sleep is lower
than when it is awake. The average power consumption for the
controller operating in this way may be less than if it were awake
all of the time.
Some techniques to reduce power in the controller during periods of
inactivity include running the controller's internal circuitry at a
lower frequency and/or lower voltage than when it is fully awake,
executing fewer instructions per unit time than when it is fully
awake, and selectively disabling some or all of the functions
within the controller.
In some embodiments, operating power may be reduced during periods
of reduced activity. For example, the controller may have several
different reduced-power modes of operation, as compared with its
full-power, full-operation mode. Often, only a subset of the full
capabilities of the controller is needed at any particular time. In
this case, some of the functions of the controller may be unused
except during specific times and may be disabled the rest of the
time to reduce average power consumption. Additionally, some of the
functions within the controller may be needed but only at a slower
operating rate so operating them at a lower frequency may reduce
average power consumption. Continuously adjusting the operating
frequency of one or more subsystems to better match the needed
performance at any particular time may also reduce the controller's
average power consumption. By automatically selecting the
appropriate low-power operating mode appropriate for the tasks it
needs to perform at any particular period of time, the controller
may reduce its average power consumption while still allowing the
SPD to operate acceptably well.
In some embodiments, operating power is reduced by running at a
lower or lowest voltage. Often, the circuitry that may be used to
implement a particular controller embodiment is specified to
operate properly over a range of voltages. Typically, the lowest
power consumption is achieved at the lowest supply voltage. In some
embodiments, the maximum frequency of guaranteed operation is
reduced when the supply voltage is reduced, even if the supply
voltage remains within the specified operating range. Also, in some
embodiments, certain functions of the controller may be guaranteed
to operate only at a higher minimum voltage than the minimum
voltage required by other functions.
By allowing the operating voltage of the controller to be varied,
some embodiments may have the controller select its operating
voltage based on the frequency at which it needs to operate and/or
which functions it needs to use during any particular period of
time. Using this technique, the average power consumption of the
controller may be lower than if the controller were to be operated
at the highest minimum guaranteed operating voltage of all of its
frequencies and functions.
In some embodiments of the SPD, updating the controller's
non-volatile program and/or data memory may allow the controller to
maintain the contents of its memory after a loss of power. In some
cases, these non-volatile memory circuits require a higher
operating voltage during update operations than during read
operations. Allowing the controller to adjust the operating voltage
of the non-volatile memory subsystem, such that it is operated at
the higher voltage only during update operations, and operated at
the lower voltage during read operations, may reduce the average
power consumption of some embodiments of the SPD.
In some embodiments, the total extra power consumption needed in
the controller when the higher voltage is used for writing to
non-volatile memory circuits may be further reduced by grouping
together the non-volatile-memory update operations. The "overhead"
time spent raising and lowering the voltage may, therefore, be
shared among multiple update operations. This means that for a
given set of update operations, the time (and excess power) spent
on most of the "overhead" may be eliminated.
A variety of methods may be used to initialize the SPD operating
parameters. One method that may be used is through the inclusion of
a USB port or similar serial interface for initial configuration
and/or initial power charge of the SPD and to use a laptop or
computer that is able to directly operate at the MAC layer of the
Wi-Fi protocol to set up the higher layer operations. In many
environments where the SPD is operating, the DHCP protocol is often
available, so the IP address(es) may be obtained automatically. In
environments where the DHCP protocol is not available or is
otherwise inappropriate to be used, the IP address(es) and other
relevant operating parameters may be allocated manually.
Some embodiments include a button on the back of the SPD that
invokes the display of the IP address momentarily. With the IP
address known, a configuration utility downloadable from a web site
or other convenient location may be used to configure the operating
behaviors of the SPD. Alternatively, for embodiments in which the
SPD includes touch-screen capability, triggering a configuration
mode may lead to a series of menu queries that may adequately drive
initial configurations.
For some of SPD applications, compromise of the host network by the
SPD is typically not possible due to the limited network interface
power budget and/or lack of standard higher-layer protocols that
will be present on the device. For example, in many SPD
embodiments, there will not typically be support for TCP
operations. The SPD functional behaviors, therefore, will be
governed by an entirely proprietary message set. In some
embodiments, however, passwords and/or other means may be used to
limit access to the SPD to authorized personnel and/or designated
remote device(s).
Referring to FIG. 12, a method (or process) 700 is shown. The
method 700 illustrates an example of a low power activation signal
monitoring process. The method 700 generally comprises a step (or
state) 702, a step (or state) 704, a decision step (or state) 706,
a step (or state) 708, a step (or state) 710, a step (or state)
712, a decision step (or state) 714, and a step (or state) 716. The
step 702 may start a low power mode cycle (or process). Next, the
step 704 may collect and store energy. Next, the step 706 may
determine the amount of energy needed for the predicted power
usage. If there is not enough energy stored, the method 700 moves
to the step 708. To reduce power usage, the step 708 may slow down
an internal clock to the CPU 202a and/or may lower the operating
voltage. If there is enough energy for the predicted power needs,
the method 700 moves to the step 710. The step 710 senses one or
more activation signals. If an activation signal is present, such
as a signal received from the step 712, the decision step 714
determines if an activation condition has been met. If not, the
method 700 moves back to the step 702. If an activation condition
has been met, the method 700 moves to the step 716. The step 716
starts a data collection mode (to be described in more detail in
connection with FIG. 13).
Referring to FIG. 13, a method (or process) 800 is shown. The
method 800 illustrates an example of a power-aware data collection
process. The method 800 comprises a step (or state) 802, a step (or
state) 804, a decision step (or state) 806, a step (or state) 808,
a step (or state) 810, a step (or state) 812, a step (or state)
814, a step (or state) 816, and a step (or state) 818. The step 802
starts the data collection mode cycle. Next, the step 804 collects
and stores energy. Next, the step 806 determines if enough energy
has been stored to perform a minimum predicted operation based on
the predicted power needs. If not, the method 800 moves to the step
808. The step 808 may perform a number of prioritization rules
based on a power budget. For example, the step 808 may
activate/deactivate one or more components, adjust the duty cycle
and/or timing interval, adjust a voltage on the components or other
power saving parameters (or rules). If there is enough energy
stored for the predicted power needs, the method 800 moves to the
state 810. The decision state 810 determines if the conditions are
still met to remain in the data collection mode. If not, the method
800 moves to the state 812, which returns to the lower power mode.
If the conditions are still met, the method 800 moves to the state
814. The state 814 stores and processes sensor data. Next, the
decision state 816 determines if the communication conditions have
been met. If not, the method 800 moves back to the state 802. If
the communication conditions have been met, the method 800 moves to
the state 818. The state 818 starts the communication mode (to be
described in more detail in connection with FIG. 14).
Referring to FIG. 14, a method (or process) 900 is shown. The
method illustrates a power-aware communications mode. The method
900 generally comprises a step (or state) 902, a step (or state)
904, a step (or state) 906, a step (or state) 908, a decision step
(or state) 910, a decision step (or state) 912, a step (or state)
914, a step (or state) 916, and a step (or state) 918. The state
902 starts the communication mode. Next, the state 904 activates
the radio (or wireless transceiver). Next, the state 906 adjusts
the broadcast power of the radio based on a power budget,
interference levels, and/or communication prioritization rules.
Next, the method 900 moves to the state 908. The state 908 sends
collected data. Next, the decision state 910 determines if an
acknowledgment has been received. If not, the method 900 moves to
the decision state 912. The decision state 912 determines if the
data received is priority data. Priority data may be considered
data that may break low power operation rules. If so, the method
900 moves back to the state 906. If the data is not priority data,
the method 900 moves to the state 914. In the decision state 910,
if an acknowledgment has been received, the method 900 moves to the
state 914. In the state 914, the method 900 receives instructions
related to predicted power needs, activation condition rules,
and/or communication rules. Next, the method 900 moves to the state
916, which sends an acknowledgment. Next the state 918 returns to
the data collection mode (e.g., the process 800).
The controller in the SPD may maintain the physical and logical
address of the SPD. A default set of these addresses may be
pre-programmed when the SPD is manufactured. If necessary, some or
all of these addresses may be set or changed when the SPD is
initially deployed and/or changed while the SPD is operational.
In the event that the SPD is deployed in a network that does not
broadcast its security information, the network interface access
codes will need to be entered as part of the initial configuration
of the SPD.
The issues of privacy and encryption are of concern when there is a
need for protection of information communicated across an open
network interface. This need may be addressed in a number of ways
with respect to SPD communications with a remote device, including:
Information communicated across the external computer network may
be sent in a proprietary format, minimizing the risk of casual
observance through the use of a network "sniffer" or similar
device. Standard encryption techniques may be applied to
communications across the external computer network for even more
robust privacy and protection against unauthorized observation.
In embodiments that include communication with the remote
device(s), this communication typically takes place through the
external computer network. Some embodiments of the SPD may contain
one or more network interfaces that implement a wireless
communications protocol. The wireless protocol may be a standard,
commonly-deployed technology, such as "Wi-Fi" (IEEE 802.11a, IEEE
802.11b, IEEE 802.11g, etc.), or the protocol may be a specialized
one particularly tailored to the low-power transmission distance,
and/or other needs of the SPD and/or its application. Combinations
of different network interface types are also possible, allowing
for gateway operations where applicable.
Regardless of the type of wireless protocol, one (or more) devices
in the external computer network typically provide the wireless
connection to the overall external computer network. Whether the
external computer network is inherently wireless or a hybrid of
wired and wireless networking technologies, the device that
operates as the intermediary to provide the connection is often
referred to as a wireless access point. In some cases, the wireless
access point may be built into a node on the network, and/or may be
part of one or more remote device(s). In some cases, the wireless
access point may be a device dedicated to bridging the wireless
network to a wired network. In some cases, the remote device may
use a built-in wireless interface to directly communicate to the
SPD (in Wi-Fi nomenclature, this is known as "ad-hoc" operation
mode).
The wireless access point used by the SPD may be a standard
commercially-available device that is typically used with other
standard wireless products or the access point may be a specialized
device whose wireless port may be used only with the SPD. The
access point may be configured to support one or more SPDs, one or
more other commercially-available wireless products, or any
combination of these. In most cases, the function of the access
point for the SPD may be to convert the wireless messages from the
SPD to the wired networking protocol used by the external computer
network and vice-versa.
A role of the external computer network when used with the SPD is
typically to transfer messages from the SPD to the remote device(s)
and vice-versa. In many cases, it is expected that standard,
already-installed computer networks may be used for this function.
Therefore, the SPD may typically be operated such that its messages
may be easily converted by the existing wireless access point(s)
for carriage across the computer network.
Typically, the SPD does not require any protocol changes in the
wired part of the external computer network and in some cases,
requires no special protocol changes for the wireless portion
either. SPD messages may therefore be constructed, formatted, and
addressed for easy transport across the external computer network
to the remote device(s), and vice-versa.
In cases where proprietary communication methods are used between
SPDs, a gateway function may exist either among the SPDs or in a
special-purpose device co-located with the SPDs but connected to
the external computer network via a standards-oriented wired or
wireless means.
Once the network connectivity has been established between the SPD
and the remote device(s), they may typically communicate with each
other using the standard messaging format and protocol(s) used in
the external computer network. Each SPD may have a unique network
address, or other means to be addressed uniquely, and likewise,
each remote device may have a unique network address, or a means to
be addressed uniquely. Properly-addressed messages may be sent
across the external computer network by both the SPD(s) and the
remote device(s) using the existing networking mechanisms.
Referring to FIG. 15, a context diagram of a system 1000 showing
wireless signal configurations is shown. The memory 204a may
contain instructions outlining a set of threshold conditions for
storing sensed data (e.g., the process 800). The memory 204a may
also contain instructions outlining a set of threshold conditions
for transmitting sensed data (e.g., the process 900). The
instructions for storing and/or transmitting data may be based on a
power budget and/or available power. The instructions may contain
computational logic for computing a threshold condition based on
one or more types of sensed data and/or one or more data points
over time. A beacon 1002 is shown connected between the network
computing services block 174 and the wireless sensing device 102.
The beacon 1002 may be used to send a signal from the network
computing device 174 to the sensor node 102. The beacon 1002 may
send a signal to wake up the node 102, or to prepare the node 102
to receive new instructions, etc. A software agent block 1010a is
shown in the mobile computing device 170a. Similarly, a software
agent block 1010n is shown in the computing device 170n. A software
agent block 1020 is shown in the network computing services block
174. The software processing may be distributed between the mobile
computing devices 170a-170n and the network computing services
block 174. A network data services block 1040 is shown connected to
the network computing services block 174. The network data services
block shows a number of classifications of network data services
1050a-1050n.
In one example, the computational logic may include a cascade
(e.g., one or more) of threshold conditions that may be used to
determine when additional features of the sensor node 102 will be
powered on or powered off. For example, when the sensor node 102
determines that one of the sensors (e.g., a sensor A) detects a
signal over a determined threshold, then one or more other sensors
(e.g., a sensor B and/or a sensor C) may be activated. In another
example, when the sensor node 102 determines that the sensor B in
combination with the node C are below a threshold, then the sensors
B and C may be turned off. In such an example, one of the sensors
160a-160n may be awake (or operational) continuously. Other sensors
160a-160n may only be powered on when the sensor node 102
determines that monitoring is needed (e.g., by the continuously
operated one of the sensors 160a-160n). Other input parameters may
be uploaded by the mobile computing device 170 to the sensor node
102. Such parameters may include a schedule of when sensor data
should be captured and/or stored.
The computational logic cascade may implement a power-aware
prioritization of sensor data capture and/or communications
information. For example, when power stored in the ESU circuit 214a
is low, the duty cycle of the sensor node 102 may be changed less
frequently. In one example, the logic may only receive signals from
the sensors 160a-160b when power is sufficient.
The computational logic may include logic for coalescing data, such
as only recording changes to a base state or expected state. A
computationally efficient process may be implemented. For example,
a process that only records the average value of one or more of the
sensors 160a-160n over a period of time may be used. The sensor 102
may begin to record events if a measurement received from the
sensors 160a-160n is significantly above average.
Since the sensor nodes 102a-102n are self-powered (through the
energy storage unit 214a), the power budget of the sensor nodes
102a-102n is generally kept small. By capturing additional energy
with the ECU 214a, the nodes 102a-102n may operate for long periods
of time on a single battery. The energy storage unit 214a may be
implemented as a battery or other storage device. The energy
capture unit 214a may store energy through various sources, such as
energy created through harvesting of light, vibration, thermal
variation, thermal gradient, infrared, chemical, RF sources or some
combination.
The computational logic for determining the threshold conditions
for storing and/or transmitting data may be based on currently
available power, predicted future energy harvesting opportunities
(e.g., based on forecasting the future environment), and/or
predicted energy consumption (e.g., based on applying the threshold
rules to the forecast environment).
A machine learning process that optimizes the predictive model may
be set up as a series of energy input functions and/or energy
consumption functions with scaling factors that are optimized based
on historical data for the sensor nodes 102a-102n. Based on prior
patterns, the model may operate with techniques of machine learning
optimization including, but not limited to, a least square
minimization of the difference between model predictions and actual
historical data, various gradient descent techniques, etc.
In order to adapt the duty cycle, the sensor sampling rates, the
adaptive and conditional powering of additional sensors based on a
set of initial sensors, the data storage and data transmission
policies of the sensor node 102, each energy use in the system 150
may be characterized as an energy cost function. For example,
energy harvesting may be a negative cost, or energy contribution.
Each data captured may be characterized as a utility or benefit
function. An optimization may be performed to maximize the utility
within a set of cost parameters. Alternatively, optimization may be
performed to maximize utility per unit of cost.
In one example, the remote computing device 170 may be a local
device on the edge of the network. For example, the remote
computing device 170 may be implemented as a smartphone, or
personal digital assistant (PDA) type device, which in turn is
connected to the central data storage and computing facility 174.
In such an implementation, some or all of the computational logic
for determining the threshold conditions and/or rules may be
implemented on the devices 170a-170n. In such an example, only the
threshold values (or simplified rules) may need to be transmitted
to the sensor nodes 102a-102n. In one example, a subset of the
processing is done on the sensor nodes 102a-102n. If power is not
an issue, the devices 170a-170n may capture a more complete stream
of data from the local sensor nodes. The data may then be processed
and/or be evaluated locally before being sent to a central server
on the computers facility 174. Such an implementation may reduce
network traffic and/or storage needed at the central data center
174. The central data center 174 may send program instructions to
one or more of the smartphones 170a-170n for computing threshold
conditions. The program instructions may in turn be sent to one of
the sensor nodes 102a-102n.
Various sources of data may be used to generate energy consumption
and/or harvesting predictions. Such sources may include (i) a
schedule of upcoming calendar events in the neighborhood of the
sensor, (ii) the proximity to or presence of nearby people,
objects, buildings, vehicles, (iii) security or threat conditions
that increase the importance of sensing, (iv) location-based data
(such as features of the location of the sensor), (v) weather
forecasts and/or other environmental conditions, (vi) scheduled or
location-based energy harvesting opportunities, (vii) presence of
increased ambient or directed RF power, (viii) known or observed
cycles including day, week, month, season, and/or (ix) other
patterns in prior data.
Various energy harvesting technologies may be used in the energy
capture unit 214a. Examples of such technologies include
piezoelectric effects of stress and/or strain, vibration, thermal
gradient, thermal variation, solar photovoltaic, ambient light
photovoltaic, infrared, radio frequency, chemical energy, air or
fluid flow, and electromagnetic fields. Many of these effects are
context dependent and may vary with location, time, proximity to
other devices or environmental features and other factors. Being
able to predict future energy inputs based on the context of the
sensor node 102 may help to determine the current power budget.
In addition to the environmental context, the predicted power
inputs and predicted power consumption may take into account
various application rules encoded in the control logic of the
memory 204a. Such inputs may be used to determine the duty cycle
sensor node 102, the sampling rate of the sensor, the conditional
powering of different components, as well as the data storage
and/or data transmission policies.
In an example of a security application used to detect intruders,
the sensor node 102 may have a very different predicted power need
than the same presence or proximity sensor node 102 used to
optimize lighting and heating based on building or room usage. The
security application may consume power when an office is closed.
The lighting and heating application may need to be active during
the scheduled office hours. The energy harvesting opportunities may
also be different. The security application may need to store
enough power during the day to last throughout the night. The
office energy application may be able to harvest enough energy
during the day to meet most needs. The transmission policies for
the security application may be immediate during non-office hours.
The transmission policy for the energy application may be daily or
weekly. In another example, a seismic vibration sensor might have a
high sampling rate after an initial threshold event, while a
temperature sensor might be needed less frequently. For many
applications, a diminishing return on the value of the data occurs
at higher sampling rates. The particular value depends on the
application.
The predicted power needs and/or harvesting opportunities may be
computed through machine learning techniques in which the
contribution of environmental features to energy harvesting and
energy consumption are learned based on prior observations. For
example, consider an application where during a time interval
t.sub.i, and at location l, energy contribution or cost is
expressed as a linear relationship between a set of environmental
features f and energy the contribution by the following equation:
.DELTA.e(t.sub.i,l)=a.sub.1f.sub.1(t.sub.i,l)+a.sub.2f.sub.2(t.-
sub.i,l)+a.sub.3f.sub.3(t.sub.i,l)+a.sub.4f.sub.4(t.sub.i,l)
For environmental factors that cost energy, the scaling factors are
negative. For environmental conditions that add to energy
harvesting, the factor is positive. The change in energy does not
go over the storage capacity of the node. The charge does not
generally fall under a threshold without shutting down the sensor
node 102.
For a time series over all the time intervals, an expression with F
being a matrix (where each row is a time interval and each column
is a feature) may result in a total energy contribution or drain
from environmental factors of E=F.times.a
The time intervals before the current time, the actual features
and/or energy needs or contributions have already been measured and
may be considered the ground truth or training set for a predictive
model. The scaling factors may be fit using well known machine
learning techniques. The functions describing dependence of the
change in energy and features of the environment may first be
estimated with known relationships (e.g., such as solar radiation
and the contribution to a photovoltaic) so that the machine
learning is focused on scaling the relationship to the particular
sensor node rather than discovering the relationship in the first
place. The sensor node 102 may be implemented with known
components. The energy inputs and/or cost functions of such known
components should be related to the technology used. The scaling
factors are normally related to the context and/or application.
In order to create the most efficient power management in a
wireless sensor network, the machine learning techniques may be
trained with data collected on a subset of the sensor nodes and/or
for a subset of time. The power budget and/or power predictions may
be calculated at a central computer in the network computing
services block 174 for larger calculations. For calculations with
fewer features, the power budget may be calculated on the remote
computing device 170. The results of the optimization of the model
may be communicated to the sensor node 102 in the form of program
instructions and/or thresholds.
The optimization of the scaling factors of the predictive model may
also take into account the effect of the control logic and the
rules currently in use to adapt the duty cycle sensor of the node
102, the sensor sampling rate, as well as the data recording and/or
transmission policies. The functions that express the relationship
between a predicted feature of the environment or context to power
consumption can include the rules in the calculation so that the
optimization is application-specific.
In some SPD applications, the software agent(s) 1010a-1010n running
on the remote device(s) 170a-170n may be used to communicate with
the SPD through the external computer network. In some embodiments,
the communication may be automatic, with each interaction triggered
by some change of status in the remote device, the SPD, or both. In
some embodiments, the communication may be manually initiated by a
user of the remote device, such as to change the configuration or
other state of the SPD. In some embodiments, both initiation modes
may be supported.
Note that in some embodiments, the user directly interacts with the
remote device(s), such as with a keyboard and/or mouse and/or
display/monitor. In some embodiments, the user interacts indirectly
with the remote device(s), such as across a computer network and/or
additional communications link.
In some embodiments, communication is automatically initiated by
the remote device. Some examples of uses of automatic communication
initiated by the remote device may include: The SPD displays a
subset of database information: a software agent running on the
remote device periodically searches a database resident in the
remote device and/or in a computer reachable through the external
computer network. The results of the search, possibly along with
other relevant information, are formatted by the software agent and
are sent to the SPD, especially when those results have changed
since the previous update. The SPD displays meeting room
information: a software agent running on the remote device monitors
the reservations for the meeting room to which the SPD is attached.
The software agent may periodically interrogate the networked
calendar and/or reservation system (such as Microsoft Outlook,
IBM/Lotus Notes, etc.) and send the reservation information,
possibly along with other relevant information, to the SPD for
display. This application may be useful for general businesses, as
well as businesses making rooms available to third parties, such as
hotels, convention centers, etc. The SPD displays examination room
information in a hospital or doctor's office. A software agent
running on the remote device monitors the patient reservations for
the room to which the SPD is attached. The information may allow
the medical personnel to know who is in the room without having to
enter the room and disturb the patient, and without having to find
the reservation clerk and ask him/her about the status of the room.
The SPD displays news, advertising, and/or market information: a
software agent running on the remote device gathers various news,
advertising, stock quotes, headlines, or other relevant information
from internal and/or external sources (such as from the Internet).
The software agent may format the information and periodically
sends the formatted information to the SPD for display. The SPD
displays pricing or other product status information: the SPD (or a
set of SPDs) may be positioned adjacent to its (their)
corresponding retail product(s). Each SPD may be used to display
up-to-date pricing and/or other product status information in a
supermarket, retail store, warehouse, or other locations. The
software agent running on the remote device may gather data to be
displayed on each SPD from a database or other computer system,
and/or from direct user input at the remote device. The SPD is used
as a personal display for each employee/user/customer: an SPD may
be attached to each employee's cubicle or office, and may be used
to display personal information, such as where the employee is
presently, if the employee is on vacation, if the employee is
available for drop-in conferences, or other employee circumstances.
The employee may update the display by sending messages through the
software agent running on the remote device, and this information
may be sent to the SPD, possibly along with other relevant
information, such as scheduling derived from other sources. The SPD
displays room status information: an SPD may be placed in a room
used by many people, such as an area with several cubicles, a ball
room, an auditorium, or other locations. The software agent running
on the remote device may gather any messages relevant to that room,
such as general notices, announcements, or other messages, and may
merge that with relevant local data, such as time, date, the
current temperature in the room, and may send the formatted
information to the SPD for display. The SPD is used as an
annunciator: the SPD may be used to display important real-time
information needed by people in the room where the SPD is located.
For example, in a technical support or telephone call-center
application, the SPD may display the average wait for customers on
hold, the number of customers presently being served, whether the
wait time is increasing, or other relevant information. The
software agent running on the remote device may collect real-time
information from the telephone switch and combine that information
with other relevant information, such as time of day, recent
history, or up time, to generate formatted data to send to the SPD
for display. The SPD is used as an actuator: the SPD may contain
one or more motors, actuators, electrical contacts, or other
similar components which may be used to move or position some
external mechanical or electrical apparatus. The SPD may receive a
message from the remote device requesting that the mechanical or
electrical apparatus be moved, repositioned, turned on, turned off,
or otherwise adjusted. Alternatively, the SPD may receive a message
requesting the status of the mechanical or electrical apparatus,
and the SPD may respond with a message reporting that status.
In some embodiments, communication is automatically initiated by
the SPD. Some examples of uses of automatic communication initiated
by the SPD may include: The SPD used as a security sensor: the SPD
may contains sensors to detect various environmental states, such
as temperature, humidity, lighting, motion, vibration, noise,
shock, pressure, or other environmental states. For example, in a
security application, a small SPD may be attached to a window and
used to detect if the window has been opened or broken. If such a
stimulus is detected, the SPD may send a message to the remote
device for further processing. The SPD may also send non-urgent
messages periodically to confirm its continued correct operation
and/or that the communication link is still functional. SPD reports
its own health status: the SPD may send messages to the remote
device to indicate that the SPD continues to operate correctly
and/or the communication link is still intact. When the SPD detects
an internal problem, such as insufficient input power and/or
internal energy storage to continue operation, the SPD may send a
message to the remote device to request maintenance service or
other appropriate action. When the remote device does not receive a
periodic health status message from the SPD, the remote device may
conclude that the communications link and/or the SPD have failed,
and the remote device may then take an appropriate action. SPD used
as a HVAC sensor: the SPD may be used in a
heating/ventilation/air-conditioning application, and may sense
temperature, humidity, lighting, or other environmental states. The
information may be sent to the remote device for further processing
with data from other sensors, and the state of the HVAC system may
then be sent to the SPD for display, along with current time, date,
local temperature, or other relevant information. SPD used as an
intercom: the SPD may contain a microphone and speaker, and perhaps
"doorbell" and "talk" buttons. The SPD may send a message to the
remote device when the "doorbell" button is pushed, and may
digitize/decode voice signals for use with the internal microphone
and speaker. The digitized voice data may be sent to and from the
remote device, or another computer interconnected with the external
computer network.
The software agent (SA) running on the remote device(s) may perform
a variety of tasks, depending upon the embodiment(s) of the SPD(s)
with which it communicates and the intended functionality of the SA
and its embodiment(s). SA embodiments typically contain
combinations of subprograms which each perform a specific task or
set of tasks. For example, some subprograms may be responsible for
reading information from a database while other subprograms write
to a database. Some subprograms may perform both tasks. SAs that
contain both readers and writers may have complex interactions with
their corresponding database(s) as well as when communicating with
the associated SPD(s).
An example SA application and associated environments that may be
used with SPD(s) are described below. Actual embodiments of SAs may
implement one, some, all, or none of the following examples.
In an exemplary embodiment, the SA may implement a calendar. This
application allows an SPD (or set of SPDs) to be used for resource
management, such as for reservation management of a conference
room. The SPD(s) used for this application may typically contain a
display of some sort that may show resource scheduling information.
The SPD(s) may communicate with the remote device(s) running the
calendar SA(s), and the calendar SA(s) may typically interact with
a separate calendar management program/database, such as Microsoft
Outlook or IBM Lotus Notes. Through the interaction with the
separate calendar management program/database, the SA(s) may gather
the necessary data to send to the SPD(s) for appropriate display of
scheduling information, and/or gather the necessary data from the
SPD(s) to appropriately update the separate calendar management
program/database.
The calendar application may be implemented in a variety of
environments. Exemplary environments include small offices using
off-site hosted MS-Exchange services, small offices using off-site
hosted World Wide Web Distributed Authoring and Versioning or
WebDAV services (a set of extensions to the Hypertext Transfer
Protocol (HTTP) that allows users to collaboratively edit and
manage files on remote web servers), medium-to-large offices using
MS-Exchange, and medium-to-large offices using local WebDAV.
For small offices using off-site hosted MS-Exchange, FIG. 16A
illustrates an expected environment for a small office using a
hosted MS-Exchange server for their mail services. In this example,
company A may have 3 conference rooms, and computer 2 may be an
administrative computer that is running Outlook and the SPD A
Software Agent (SA). Note that in this example, multiple companies
share a single MS-Exchange server, accessing the server via their
internet connection perhaps over SSL or VPN connections through a
firewall. Routers, hubs, etc. are assumed to be in the LAN cloud
but are omitted for clarity.
The configuration illustrated in FIG. 16A poses several challenges:
Installation/operational simplicity: Small offices often operate
with little to no information technology staff available for
applications training, management or installation. Exchange server
modification avoidance: hosting companies providing MS-Exchange
services often resist extensions to shared server platforms.
Additional outlook license avoidance: MS-Outlook seat license costs
are often high for small offices. As a consequence, they typically
have less volume discount and purchasing leverage. Minimizing
communications hops: Offsite hosting often creates large, variable
latencies in communications to the MS-Exchange server. A variety of
software architectures may be used to address these challenges.
These are described below.
In some embodiments, installation and operational simplicity are
linked as an issue but may be accomplished in several only
somewhat-related ways as follows: Software Agents (SAs) may be
designed to be installed only on client workstations.
In this way, access to a server machine by an SPD administrator may
not be necessary. Two software elements may be used to fulfill the
SA behaviors in each of two client workstations: one is an Outlook
VBA extension (a version of Visual Basic that is embedded in
Outlook) and the other is a separate supporting service. Both
software elements may be installed in the same process. As a
consequence, decisions about which is needed under what conditions
may not be necessary. Delegation may be "hooked" so that the
notification of delegate status for a resource managed with an SPD
may be provided to the (potentially new) delegate with the option
of installing the necessary components to manage the resource.
Standard Outlook resource management may be used for creation and
management (i.e., delegation) of the resource as a mailbox. This
may allow meeting organizers, even for those individuals who are
not administrators of the rooms, to use the source when scheduling
meetings. Standard Outlook resource management may be used to avoid
data collisions or conflicts in accepting resource booking
requests. Installation of VBA elements may be largely invisible. To
all appearances there may be an additional tool available to manage
the SPDs and client software but otherwise it is just standard
Outlook for handling the resource. The supporting service may be
"fed" information by the Outlook VBA extension in the background on
an event-driven basis. This may provide display updates to the SPD,
such that no user intervention may be necessary. The supporting
service may be a member of a cooperating group of services running
on the machines of all resources owners and delegates. SPD
communication may be performed only by the currently elected master
whose role may be constantly negotiated behind the scenes without
administrator input.
In some embodiments, utilization of the Outlook VBA mechanism and a
local high-priority software service may allow scheduling and
creation of resources without changes to the MS-Exchange
server.
In some embodiments, utilization of the Outlook VBA mechanism and
the creation of the resources as a resource mailbox may avoid any
need for an additional Outlook license.
In some embodiments, utilizing the locally-located workstation of
the administrator(s) reduces the latency and communications delays
to a best case path over the LAN versus the uncertain path to the
MS-Exchange server, thereby minimizing communication hops.
For small offices using off-site hosted WebDAV, FIG. 16B
illustrates an environment for a small office using an offsite
hosted iCalendar (a calendar application developed by Apple
Computer). In this example, company A may have 3 conference rooms,
and computers 1 and 2 may be running the SPD management utilities
and Software Agent (SA). Note that the server is accessed via an
Internet connection, such as over SSL or VPN connections through a
firewall. Routers, hubs, etc. are assumed in the LAN bubble above
but are omitted for clarity.
The configuration illustrated in FIG. 16B poses several challenges:
Installation/operational simplicity: Small offices often operate
with little to no information technology staff available for
applications training, management or installation. The generic
nature of WebDAV/iCal services: No calendar behaviors really exist
within a WebDAV/iCal service. Instead, the application provides a
web-based transactional-based information storage service on which
a group calendaring system may be built. Minimizing communications
hops: Offsite hosting creates large, variable latencies during
communication with the WebDAV server. A variety of software
architectures may be used to address these challenges. These are
described below.
In some embodiments, installation and operational simplicity are
linked as an issue but are accomplished in several only somewhat
related ways as follows: Installation of software may be limited to
client workstations such that no administrator access to any form
of server machine is necessary. Two software elements may be used
to fulfill the SA behaviors in each of two client workstations. One
of these may be an SPD management utility to fulfill the same
purpose as the Outlook VBA application. The other may be a separate
supporting service. Both software elements may be installed in the
same process. In this way, no decisions regarding which one is
needed under what conditions may be necessary. Delegation may be
handled using an email notification that indicates the delegation
status for the resource to be managed. An attached executable file,
which is generated and pre-configured for the WebDAV options
necessary, may be optionally installed to install the components
necessary to manage the resource.
For medium-to-large offices using MS Exchange, FIG. 16C illustrates
a local MS-Exchange server for their mail services. In this
example, Company A may have 3 conference rooms, and computers 1 and
2 may be running Outlook and the SPD Software Agent (SA). Note the
more direct connections between the system components. Once again,
routers, hubs, etc. are assumed in the LAN bubble above but are
omitted for clarity.
For medium-to-large offices using off-site hosted WebDAV, FIG. 16D
illustrates a local WebDAV server for iCalendar services. In this
example, company A may have 3 conference rooms, and computers 1 and
2 may be running SPD management utilities and Software Agent (SA).
Note the more direct connections between the system components.
Once again, routers, hubs, etc. are assumed in the LAN bubble above
but are omitted for clarity.
In other embodiments (not shown), the SPDs may be used in systems
that implement or use Microsoft Outlook and/or Lotus Notes (an
application offered by IBM).
The remote device(s), such as computer devices 104 (e.g., as
described in FIG. 1) may have a CPU and a memory (e.g., shown as
122 and 120 in FIG. 1, respectively). The CPU of the remote
device(s) 104 may be similar to the CPU 202 of the SPDs 102. The
memory of the remote device(s) 104 may be similar to the memory 204
of the SPDs 102. The type, functionality, throughput and/or
capacity of the CPU and/or the memory of the remote device(s) 104
may be varied according to the design criteria of a particular
implementation.
The remote device(s) 104 may also be implemented as the mobile
computing device 170 and/or the network data services device 172.
For example, the software agent 1010a may be stored in the memory
of the mobile computing device 170a (e.g., as shown in FIG. 15). In
another example, the software agent 1020 may be stored in the
memory of the network computing services 174 (e.g., as shown in
FIG. 15). The implementation of the remote device(s) 104 may be
varied according to the design criteria of a particular
implementation.
Effective use of equipment and/or resources may be an important
factor for management of an enterprise environment (e.g., a
hospital, a corporate office, a government services building,
etc.). For example, efficient room reservation may prevent
conflicting use of resources. In some embodiments, pieces of
equipment (e.g., resources) may not be fixed to rooms and/or
locations. For example, resources may be mounted on movable
stations and/or trolleys. In another example, some equipment may be
leased and/or rented for a particular occasion.
In an embodiment implemented in a hospital, room reservations may
be expanded to include movable equipment. Generally, enterprise
environments may want an effective way to trace/track resources to
ensure resources are available and not underutilized (e.g., used
inefficiently). Different types of rooms may have different
purposes. For example, some rooms may be used for consultations
and/or examinations. In another example, some smaller rooms may be
for patient care and/or small meetings. In yet another example,
rooms may be used for operations (e.g., a sterile environment). In
another example, large rooms may be used for large gatherings
(e.g., dozens and/or hundreds of people). In still another example,
some rooms may have expensive lab equipment such as MRIs and/or
X-ray machines. The predictability of needs for each type of room
may be different. There may be an associated risk when a piece of
equipment and/or resource goes offline and/or becomes
unavailable.
Resources may be rooms, equipment (e.g., stationary equipment such
as MRI machines, moveable equipment such as defibrillator units,
etc.), personnel, heating and cooling systems, emergency systems,
security systems, etc. The type of resource may be varied according
to the design criteria of a particular implementation. The
resources may have a consumable such as a refillable product (e.g.,
toner, water level, chemical level, power capacity, etc.). For
example, if the resource is personnel the consumable may be
considered an amount of time available for efficient and/or safe
work before rest is needed. In some embodiments, the resources may
have an expiry time (e.g., a time when the consumable needs
replacing, a time before maintenance is needed, a time before the
entire resource should be replaced, etc.). The resources should be
configured to satisfy health, safety and/or regulatory standards.
The type of resource and/or consumable may be varied according to
the design criteria of a particular implementation.
In some embodiments a resource may be backed-up with a secondary
power source (e.g., a battery). For example, in regular
circumstances the battery may not be used, but the battery may have
a mandated level of charge to ensure compliance (e.g., safety
concerns, regulatory compliance, etc.). The level of charge of the
batter may be affected by storage time, battery deterioration
and/or actual usage. For example, battery-powered resources,
especially movable equipment, should satisfy battery
compliance.
Generally, enterprise environments want the right equipment at the
right place to ensure efficient use of resources. For example, in
some cutting-edge hospitals stations may be implemented as mobile
robots. In another example of a conventional hospital environment,
hospital stations may flag and/or signal an orderly, nurse and/or
technician to indicate where to be and/or go next. The remote
device(s) 104 may implement scheduling data to ensure efficient use
of resources.
Determining scheduling data may be a resource-constrained project
scheduling problem (e.g., RCPSP). The remote device(s) 104 may
formulate the RCPSP to determine the scheduling data. The
scheduling data may be comprised of various activities. The remote
device(s) 104 may determine that there are N activities in a
project. For example, a set (e.g., J={j_0, j_1, . . . , j_N,
j_N+1}) may denote the set of activities. The remote device(s) 104
may implement restraints to manage and/or solve the RCPSP. For
example, constraints may be that activities cannot be both
segmented and preempted (e.g., the activities have to be
scheduled).
Generally, some activities may have to precede others (e.g., have a
higher priority, be completed sooner, etc.). The remote device(s)
104 may determine precedence and/or priority by managing precedence
relationships. For example, the precedence relationships may be
represented by an acyclic graph (e.g., represented as G=(V, E))
stored in the memory of the remote device(s) 104. The vertices V of
the acyclic graph may be a set of activities. The edges E of the
acyclic graph may represent the precedence relationship. In some
embodiments, a placeholder activity (e.g., j_0) having no
predecessor and/or a placeholder activity (e.g., j_N+1) having no
successor may be implemented for analytical convenience.
The remote device(s) 104 may determine a set (e.g., Q={q_0, q_1, .
. . , q_i, . . . , q_M}) to denote renewable resource types. The
remote device(s) 104 may determine a set (e.g., R={rq_0, rq_1, . .
. , rq_i, . . . , rq_M}) to denote a resource capacity for resource
of type. For example, the value rq_i may be a value greater than or
equal to zero indicating the resource capacity of the renewable
resource type q_i. Each activity in the set J may have an
associated duration (e.g., dj_0, dj_1, . . . , dj_i, . . . ,
dj_N+1). Each activity in the set J may have an associated resource
requirement (e.g., rjq_0, rjq_1, . . . , rjq_i, . . . , rjq_M),
where rjq_i may represent a requirement of resource the q_i when
activity j is scheduled.
The remote device(s) 104 may store a schedule. The schedule may be
denoted by a vector (e.g., S={sj_0, sj_1, . . . , sj_N, sj_N+1}),
where sj_N denotes a start time of the activity j_N. A makespan may
be defined. The makespan may be a difference between a maximum
finish time and a minimum starting time over all activities. The
remote device(s) 104 may determine whether a schedule is feasible.
For example, the schedule may be feasible if (1) no activity is
started before predecessors of the activity are finished and (2)
for every time t, a sum of the resource requirements for all
ongoing activities does not exceed enforced capacities.
The RCPSP solved by the remote device(s) 104 may comprise finding a
feasible schedule with a minimal makespan. The remote device(s) 104
may implement techniques to solve the RCPSP. For example, one
technique to solve the RCPSP is the Ant Colony Optimization method.
More details on Ant Colony Optimization may be found in "Ant Colony
Optimization for Resource-Constrained Project Scheduling", by
Merkle et al, IEEE Trans. On Evolutionary Computation, Vol. 6,
August 2002 and "Using an Enhanced Ant Colony System to Solve
Resource-Constrained Project Scheduling Problem", by Chen et al,
IJCSNS, Vol. 6, No. 11, November 2006, the appropriate sections of
which are incorporated by reference. The technique implemented to
solve the RCPSP may be varied according to the design criteria of a
particular implementation.
The remote device(s) 104 may determine the scheduling data by
mapping the management problem to the RCPSP. In some embodiments,
rooms, movable equipment stations and/or personnel may be denoted
as resources. Resources of the same type may be considered
fungible. For an example in a hospital, there may be 5 examination
rooms which are all similar and/or identical for most purposes. The
5 examination rooms may be given a resource label of `Examination
Room`, and a resource capacity value of 5. In another example,
there may be 3 identical rooms for meetings. The identical meeting
rooms may be given a resource label of `Customer Meeting Room`
and/or `Staff Meeting Room`.
Some resources may be non-fungible for some purposes. For an
example in a hospital, there may be four ophthalmologists on call,
all fungible for most purposes (e.g., have the same credentials
and/or capable of performing the same activities). However, if a
patient requests a specific specialist then the requested
ophthalmologist become non-fungible (e.g., the resources may not be
capable of performing the same activity). In another example,
various meeting rooms may be suitable for most purposes and may be
part of the same resource capacity. However, one room may have
extra equipment (e.g., a projector) and may be non-fungible for a
particular purpose. The remote device(s) 104 may label each
resource having a non-fungible purpose with a unique label. The
unique label may be one implementation to resolve non-fungible
functions performed by the resources.
Customer and/or patient services and/or demands may be mapped as
activities. In one example, if a patient requests a consultation
with a dermatologist then the consultation may become an activity
to be scheduled. In another example, if a patient is to be admitted
overnight then the room reservation may become an activity. In yet
another example, if a client sets up a meeting with a facility
manager then the facility manager meeting may become an activity.
In still another example, if a group books an outdoor activity on
the front lawn of an office complex over the weekend, the booking
of the front lawn may become an activity.
The activities may be assumed to last a pre-determined time unit
(e.g., a time slot of one hour, a time slot of 30 minutes, a time
slot of 15 minutes, etc.). Activities that last longer than the
pre-determined time slots may be broken-up into units having the
same time as the pre-determined time slots. Activities that last
longer than the pre-determined time slots may form contiguous
blocks of time. Contiguous blocks of time may be enforced by
storing these as precedence relationships in the acyclic graph
G=(V, E).
In some embodiments, specific treatments and/or studies during a
patient stay may be represented as activities. For example, a
heart-surgery patient may need a lung capacity test twice a day,
which may be entered as two activities. The precedence
relationships may be that both of the lung capacity tests have to
be scheduled after the surgery. Further precedence relationships
may be that the second of the lung capacity tests should be
performed after a certain amount of time after the first of the
lung capacity tests.
The remote device(s) 104 may model the enterprise management as a
type of RCPSP. For an example of a hospital, the hospital may want
to clear an activity queue as quickly as possible (e.g., to
discharge patients quickly and/or file claims with the insurance
company). However, unlike a conventional RCPSP, the scheduling data
in the enterprise environment may be revised constantly (e.g., at
pre-defined intervals, as new information is received, as prompted
by a user, etc.).
The remote device(s) 104 may perform refinement rounds for various
time frames (e.g., similar to how weather forecasting operates).
For example, an RCPSP refinement round may be performed by the
remote device(s) 104 for the next 6-8 hours. In another example, an
RCPSP refinement round may be performed by the remote device(s) 104
for the next 24 hours. In yet another example, an RCPSP refinement
round may be performed by the remote device(s) 104 for the next 5-7
days. The time frame of the refinement rounds may be varied
according to the design criteria of a particular
implementation.
The SPDs 102 may be assigned and/or paired with every possible
resource. Assigning the SPDs 102 to each resource provides each
resource with an ability to self-monitor, switch to stand-by mode
and/or wake from sleep when the resource is scheduled to be used.
Using the SPDs 102, the resource may communicate with the remote
device(s) 104. The SPDs 102 may select one or more modes based on
the scheduling data. For example, the SPDs 102 may enter an active
state during an activity and/or an inactive (e.g., low-power) mode
during an activity based on whether the assigned resource is used
for the particular activity.
In some embodiments, personnel may wear one of the SPDs 102. For
example, in a hospital environment emergency residents may be
assigned one of the SPDs 102. The SPDs may measure an alertness
(e.g., based on hours spent working) of the emergency workers. The
SPDs 102 may indicate to the emergency workers that time is
available to rest. While the emergency workers are resting, the
SPDs may communicate to the remote device(s) 104 that the worker is
unavailable. In another example, the SPDs 102 may indicate that
unionized employees should stop working to prevent the employees
from exceeding overtime regulations.
The SPDs 102 may be in communication with the remote device(s) 104.
The SPDs 102 may store updated the scheduling data communicated by
the remote device(s) 104 in the memory 204. For example, the SPDs
may communicate periodically, as new information is available to
send, and/or at pre-determined time periods. The remote device(s)
104 may receive updates about incoming activities involving the
resources assigned to the SPDs 102. For example, a mobile
instrument station may start signaling and/or flagging that the
resource (e.g., the mobile instrument station) should be moved to a
new location (e.g., to be used for a different activity). In
another example, a consultation room may display a name of the next
patient on a door plate display (e.g., one of the SPDs 102
configured to operate as a display). In yet another example, an ICU
nurse may be messaged by the SPD 102 to perform a check (e.g., one
of the activities). In still another example, a staff member may be
reminded by the SPD 102 to check that a conference room is prepared
for a meeting. The remote device(s) 104 may send updated scheduling
data to the SPDs 102 (e.g., the scheduling data may be updated
based on information sent by the SPDs 102 to the remote device
104).
The SPDs 102 may be configured to flag the assigned resource as
unavailable. For example, the SPD 102 may perform a diagnostics
test to detect a malfunction (e.g., of the assigned resource and/or
the SPD 102). If the SPD 102 flags the resource as unavailable a
communication may be sent to the remote device(s) 104. The remote
device(s) 104 may reduce the resource capacity of the resource that
is unavailable. For example, the next refinement round may not
include the unavailable resource. The unavailable resource may not
be used during the refinement rounds until the SPD 102 communicates
that the resource is available (e.g., repaired, refilled,
recharged, etc.).
In some embodiments, the SPDs 102 may be notified of future
activities based on the scheduling data. The SPDs 102 may
communicate to the remote device(s) 104 that the assigned resource
may be depleted soon (e.g., the remote device 104 may check the
power budget of the SPD 102 and/or the assigned resource). In one
example, a portable generator may have enough gasoline (e.g., gas
may be a consumable of the portable generator resource) for a
current outdoor event, but not the next event. The resource
assigned to the SPD 102 may be capable of completing the activity
currently assigned but may not be capable of completing upcoming
activities. For example, the back-up battery of the assigned
resource may have enough charge for the ongoing shift, but not
enough charge for the next day. The remote device 104 may
communicate to the SPD 102 to enter a conservation and/or recovery
mode of operation when the SPD 102 has an insufficient power
budget.
The SPD 102 may signal to the centralized RCPSP (e.g., the remote
device(s) 104) that the assigned resource will be unavailable. The
remote device(s) 104 may retire the resource from the refinement
rounds for the next day, but not from the refinement rounds for the
current shift. The resource paired with the SPD 102 may run a local
scheduling problem to opportunistically go into recharge and/or
maintenance mode when not expected to be in use. For an example of
irrigating an outdoor lawn, there may be a maintenance operation
scheduled for a particular time of day (e.g., 8 pm). If another
activity (e.g., an outdoor corporate function) spans past 8 pm the
maintenance mode may not start. The SPDs handling the sprinkler
system may delay irrigation to the next available opportunity.
The enterprise environment may also comprise an outdoor environment
(e.g., the front lawn of an office complex). For example, the SPDs
102 may measure grass deterioration. If the grass deterioration
level exceeds a pre-determined amount, the resource (e.g., use of
the lawn) may be retired until the grass recovers.
The remote device(s) may include risk metrics when determining the
scheduling data. The resources may be assigned usage risk (e.g.,
based on a current condition of the resource, a capacity of the
resource, etc.). Each time slot may be assigned a risk capacity.
For example, the risk capacity may be a risk budget the enterprise
environment is capable of handling and/or may be afforded at a
particular time. The determination of usage risk and/or the risk
capacity of time slots may be varied according to the design
criteria of a particular implementation.
For an example of a hospital, fewer specialists and/or personnel
may be available during a night shift. The remote device(s) 104 may
determine there is low risk capacity during the time slots of the
night shift (e.g., fewer resources are available if some resources
fail). In some embodiments context information and/or historical
data may be used to determine risk capacity. For example, the
enterprise environment may expect more activity on certain dates.
The remote device(s) 104 may be programmed (e.g., based on
management decisions of the enterprise environment) to lean towards
safety prior to and/or during a particular period and assign a low
risk capacity.
The SPDs 102 assigned to the resources may be configured to
self-monitor and/or update a risk cost associated with the assigned
resource. The SPDs 102 may determine the risk cost associated with
the assigned resource based on various factors. For example, the
risk cost associated with the resource may be inferred from
collected data and/or historical data using machine learning
techniques. In an example of a hospital, the risk cost may be a
function of a medical history of the patient and/or a patient type
(e.g., the risk cost may not always be a function of something
intrinsic to the resource). In another example, particular
employees may be known to be careless with some resources. The
resource may be assigned a higher risk cost when assigned to the
careless employee.
Based on the risk metrics the scheduling data may be different. For
example, risky activities may be spread in time to avoid clustering
risky activities together. In another example, risky activities may
be managed by scheduling the activity when more resources are
available (e.g., provide a higher probability of avoiding the risk
and/or manage the risk better if an undesirable outcome
occurs).
Referring to FIG. 18, a flow diagram of a method 1100 is shown. The
method 1100 may categorize resource availability. The method 1100
generally comprises a step (or state) 1102, a step (or state) 1104,
a decision step (or state) 1106, a decision step (or state) 1108, a
step (or state) 1110, a step (or state) 1112, a decision step (or
state) 1114, a step (or state) 1116, a decision step (or state)
1118, a step (or state) 1120, a step (or state) 1122, a step (or
state) 1124, and a step (or state) 1126.
The step 1102 may be a start state. In the step 1104, the remote
device(s) 104 may categorize one or more resources (e.g., based on
information from the SPDs 102). The decision step 1106 may
determine if there are more resources to categorize. If so, the
method 1100 moves to the decision step 1108. If not, the method
1100 moves to the step 1110. The decision step 1108 determines if a
particular resource is in the same category as another resource. If
so, the method 1100 moves to the step 1112. If not, the method 1100
moves back to the step 1104. The step 1112 may increase a resource
capacity. Next, the decision step 1114 determines if a particular
resource has a non-fungible purpose. If not, the method moves back
to the step 1104. If so, the method moves to the step 1116. The
step 1116 applies a unique label to the resource for the
non-fungible purpose.
In the step 1110 the remote device(s) 104 may determine a risk
factor associated with each of the resources (e.g., based on
information from the assigned SPDs 102). Next, the decision step
1118 determines if a resource has been flagged as unavailable. If
so, the method 1100 moves to the state 1120. If not, the method
1100 moves to the state 1122. The step 1120 marks a resource as
considered unavailable for scheduling. The method 1100 then moves
to the step 1124, which provides a resource availability data for
scheduling. The step 1122 marks a resource as considered available
for scheduling. The method 1100 then moves to the step 1124. Next,
the step 1126 ends the method 1100.
Referring to FIG. 19, a flow diagram of a method 1140 is shown. The
method 1140 may determine scheduling data. The method 1140
generally comprises a step (or state) 1142, a step (or state) 1144,
a step (or state) 1146, a decision step (or state) 1148, a step (or
state) 1150, a step (or state) 1152, a step (or state) 1154, a step
(or state) 1156, a decision step (or state) 1158, a step (or state)
1160, a step (or state) 1162, and a step (or state) 1164. The step
1142 starts the method 1140. In the step 1144, the remote device(s)
104 receives resource availability data. Next, the step 1146 maps
service demands as activities. Next, the decision step 1148
determines if the activity is longer than a time slot. If so, the
method 1140 moves to the step 1150. If not, the method 1140 moves
to the step 1152. The step 1150 schedules the activity units with a
high precedence relationship. The step 1150 moves to the step 1152.
In the step 1152, the remote device(s) 104 determines a precedence
relationship between the activities.
Next, the step 1154 determines a risk capacity of the time slots.
Next, the step 1156 schedules the activities based on precedence
relationships, risk capacity and/or resource availability data.
Next, the decision step 1158 determines if another round of
refinements are needed. If so, the method 1140 moves to the step
1160. If not, the method 1140 moves to the step 1162. The step 1160
selects a time frame of a subsequent refinement round, then moves
back to the step 1144. The step 1162 sends scheduling data to the
SPDs 102. The step 1164 ends the method 1140.
Referring to FIG. 20, a flow diagram of method 1180 is shown. The
method 1180 may operate the SPDs 102 and/or the associated
resources based on the scheduling data. The method 1180 generally
comprises a step (or state) 1182, a step (or state) 1184, a step
(or state) 1186, a decision step (or state) 1188, a step (or state)
1190, a step (or state) 1192, a step (or state) 1194, a step (or
state) 1196, a decision step (or state) 1198, a decision step (or
state) 1200, a step (or state) 1202, a step (or state) 1204, and a
step (or state) 1206. The step 1182 generally comprises a start
step. In the step 1184, the SPDs 102 cycle between various modes
based on default patterns (e.g., patterns already scheduled and/or
pre-configured patterns). Next, the step 1186 checks the scheduling
data. Next, the decision step 1188 determines if the scheduling
data indicates an activity. If so, the method 1180 moves to the
step 1190. If not, the method 1180 moves back to the state
1184.
The step 1190 turns on one or more active SPDs for a particular
activity. In the step 1192, the SPDs inactive for the particular
activity may enter a low powered mode. Next, in the step 1194, the
inactive SPDs may be flagged as available and/or the active SPDs
may be flagged as unavailable (e.g., for refinement rounds
performed by the remote device(s) 104). Next, the step 1196 checks
scheduling data. The decision step 1198 determines if a particular
activity has ended. If so, the method 1180 moves back to the step
1184. If not, the method 1180 moves to the step 1200. The decision
step 1200 determines if another activity has been scheduled. If so,
the method 1180 moves to the step 1202. If not, the method 1180
moves to the step 1204. The step 1202 adjusts a status of the SPDs
102 according to the priorities (e.g., based on the precedence
relationships). The step 1204 maintains the status of the SPDs 102.
The step 1206 ends the method 1180.
Referring to FIG. 21, a flow diagram of a method 1220 is shown. The
method 1220 may update scheduling data for the SPDs 102 from the
remote device(s) 104. The method 1220 generally comprises a step
(or state) 1222, a step (or state) 1224, a decision step (or state)
1226, a step (or state) 1228, a step (or state) 1230, a decision
step (or state) 1232, a step (or state) 1234, a step (or state)
1236, a step (or state) 1238, and a step (or state) 1240.
The step 1222 may be a start step. The step 1224 may cycle modes of
the SPDs 102 based on one or more scheduled patterns. Next, the
decision step 1226 determines if a communication mode has started.
If not, the method 1220 moves back to the step 1224. If so, the
method 1220 moves to the step 1228. In the step 1228, the SPDs 102
communicate with the remote device 104. Next, the step 1230 checks
the scheduling data. Next, the decision state 1232 determines if
the remote device 104 has updated the scheduling data. If so, the
method 1220 moves to the step 1234. If not, the method 1220 moves
to the step 1236. The step 1234 updates the SPD scheduling data.
The step 1236 ends communication with the remote device 104. The
step 1238 cycles the modes of the SPDs based on the scheduled
patterns (e.g., the scheduling data). Next, the step 1240 ends the
method 1220.
Referring to FIG. 22, a flow diagram of a method 1260 is shown. The
method 1260 may check a power budget of the SPDs 102. The method
1260 generally comprises a step (or state) 1262, a step (or state)
1264, a step (or state) 1266, a decision step (or state) 1268, a
step (or state) 1270, a decision step (or state) 1272, a step (or
state) 1274, a step (or state) 1276, and a step (or state) 1278.
The step 1261 may be a start step. Next, the step 1264 may cycle
modes of the SPDs 102 based on the scheduling data. Next, the step
1266 may check the scheduling data. Next, the decision step 1268
determines if there is an upcoming activity. If not, the method
1260 moves back to the step 1264. If so, the method 1260 moves to
the step 1270. The step 1270 checks the power budget of the SPDs
102 used for the specified activity. Next, the decision step 1272
determines if the SPDs 102 have enough power for the activity. If
so, the method moves back to the step 1264. If not, the method 1260
moves to the step 1274. The step 1274 removes the SPDs with
insufficient power from the applicable refinement rounds (e.g., the
next shift, the next day, the next week, etc.). Next, the step 1276
conserves and/or recovers power for the SPDs with insufficient
power. Next, the step 1278 ends the method 1260.
The SPDs 102 may be configured to perform a self-assessment of the
assigned/paired resource. The self-assessment may determine an
amount of measured wear of the assigned resource. For example, the
measured wear may be a fluid level of the resource (e.g., a
lubricant level). In another example, the amount of measured wear
may be a number of recharges of a rechargeable battery pack. The
SPDs 102 may communicate the amount of measured wear to the remote
device(s) 104. In some embodiments, the SPDs 102 may be configured
to run and/or solve a local scheduling problem to opportunistically
enter a maintenance mode for the assigned resource. The maintenance
mode may be entered based on the scheduling data and/or the amount
of measured wear (e.g., a function of the measured wear). In some
embodiments, the maintenance mode may be performed by the SPD 102
and/or the assigned resource (e.g., recharging a battery, cooling
down, waiting for a lawn to regrow, etc.). In other embodiments,
the maintenance mode may be performed by personnel (e.g., repair
personnel).
In some embodiments, the resources may be leased equipment. The
SPDs 102 assigned and/or paired to leased resources may flag the
leased resource as unavailable. For example, a manager of the
enterprise environment may determine a budget (e.g., a cap) for
various resources. The leased resource may have a cost driven by
actual usage. The leased resource may be flagged as unavailable if
an accumulated cost over a time period exceeds a pre-determined
cap. For example, a vehicle may be leased based on mileage (e.g.,
use). When the vehicle reaches a particular mileage, the cost of
continued use of the vehicle may exceed the available budget. The
vehicle may be flagged as unavailable to prevent exceeding the
available budget.
In some embodiments, the remote device(s) 104 may assign prices
(e.g., a representation of a value of the resource) to the
resources. The activities in the scheduling data may be assigned
budgets. The assigned prices of the resources may go up or down
based on an amount and/or a demand of the resource. For example,
the assigned price of a resource may increase as the resource
becomes depleted (e.g., has a lower capacity). In another example,
the assigned price of the resource may increase when the resource
is in high demand. The SPDs 102 and/or the remote device(s) 104 may
communicate and/or update the assigned prices of the various
resources. The remote device(s) 104 may use the assigned prices of
the resources and/or the assigned budgets of the activities to
determine the scheduling data.
In some embodiments, the SPDs 102 assigned and/or paired with the
resources may be configured to adjust settings of the assigned
resources. For example, a setting may be a high-speed mode and a
low-speed mode for a fan. In another example, the setting may be an
intensity of lights in a meeting room. In yet another example, the
setting may be turning on/off lawn sprinklers. The type of setting
may be varied according to the design criteria of a particular
implementation.
Some resources may perform similar activities. In some embodiments,
using one resource may have an indirect effect. For example, a
setting to open a ceiling skylight in an office building may be
performed to let in more natural light. However, an indirect effect
may be that opening the ceiling skylight lets more heat into the
office building. For example, the remote device(s) 104 may know
that the ceiling skylights are scheduled to be open and a heat
level in the enterprise environment may be increased. Based on the
activation of one resource (e.g., the skylight being opened)
another of the resources may have a setting changed (e.g., the air
conditioning system may be activated to operate at a higher
capacity).
In some embodiments, the SPDs 102 may be mobile SPDs. Mobile SPDs
may be moveable devices (e.g., moved manually and/or moved
autonomously). The mobile SPDs may be configured to perform tasks
and/or activities in situations where interaction with people is
limited. For example, the mobile SPDs may be implemented in remote
locations (e.g., hard to reach) and/or locations with limited
access. The mobile SPDs may be implemented in situations where
recharging options (e.g., charging the battery 214) are limited
and/or unreliable. In some embodiments, the computational resources
of the mobile SPD may be limited.
The remote device(s) 104 may be configured to manage the resource
capacity of the mobile SPDs. The resource capacity of the mobile
SPDs may be the battery 214 (e.g., the battery capacity) and/or
other energy storage medium (e.g., the collector 210). The remote
device(s) 104 may optimize/refine the scheduling data for the
activities based on the resource capacity of the mobile SPDs. The
remote device(s) 104 may monitor, coordinate and/or instruct
several (e.g., dozens or hundreds) of the mobile/movable SPDs. The
remote device(s) may be configured to operate a network of the
mobile SPDs efficiently and/or ensure the mobile SPDs remain
operational as long as possible (e.g., extend the battery life of
the mobile SPDs).
Extending the battery life of a mobile SPDs may be an important
consideration with respect to the "Internet of Things". For
example, many applications involve devices having non-replaceable
batteries and/or devices that are discarded once the battery is
depleted. In another example, when the mobile SDPs are implemented
in locations with limited access, replacing the battery 214 may
incur a large cost. In yet another example, the battery 214 may be
recharged using energy harvesting mechanisms that are not capable
of a fast recharge.
Each of the SPDs 102 may be configured to perform a set of tasks
(e.g., terminal tasks). Each of the tasks may comprise a sequence
of instructions (e.g., native instructions). The native
instructions may be part of an instruction set of the SPDs 102.
Each of the terminal tasks may have an associated expected
computation cost (e.g., in terms of battery charge depleted). To
complete the activities in the scheduling data, the SPDs 102 may
perform a set of terminal tasks (e.g., the SPDs 102 may determine a
set of terminal tasks needed to perform the activity). The RCPSP
formulation implemented by the remote device(s) 104 may include the
computation cost of the SPDs 102 involved in the performance of the
activity. For example, sending messages, performing updates and/or
information requests to the SPDs 102 may be the activities. The
remote device(s) 104 and/or the SPDs 102 may be configured to
determine the computational cost of performing the tasks.
The SPDs 102 may store the instruction set (e.g., in the memory
204). The instruction set of the SPDs 102 may be fixed and/or
updatable. Generally, the instruction set comprises lower-level
functions from the point of view of the remote device(s) 104. For
example, the activities may be considered high-level instructions
comprised of a sequence of tasks to be performed by the SPDs
102.
The SPDs 102 may perform the high-level instructions (e.g., the
activities) by executing sequences of the tasks. The tasks may be
implemented with the instructions in the instruction set. The tasks
that are capable of being executed intrinsically (e.g., the SPDs
102 are capable of executing the tasks without further instructions
and/or sequencing from an outside service) may be considered the
terminal tasks. Each terminal task may have an expected and/or
estimated battery consumption (e.g., a computational cost). For
example, a task `send current temperature reading` may have a lower
computational cost (e.g., estimated battery consumption) than a
task `capture video feed for 5 minutes, compress the video and
communicate the video`. The types of tasks implemented and/or the
computational cost associated may be varied according to the design
criteria of a particular implementation.
The computational cost of the tasks may be mapped in terms of
battery discharge. In some embodiments, the SPDs 102 may store the
computational cost of the tasks. In other embodiments, the remote
device(s) 104 may store the computational cost of the tasks. For
example, the mapping of the tasks to the computational cost may be
provided by a firmware engineer, a programmer of the SPDs 102
and/or a compiler. In some embodiments, the mapping of the tasks to
the computational cost may be estimated based on historical data
using machine learning techniques implemented by the remote
device(s) 104 (e.g., context information).
The SPDs associated (e.g., in an active mode) for the activity may
execute one or more of the tasks to perform the activity. The
activity may correspond to a set of terminal tasks executed either
in sequence, parallel and/or any combination thereof. In some
embodiments, the computational cost (e.g., in terms of battery
usage) may be a sum of the computational costs for each of the
terminal tasks executed to perform the activity. The computational
cost may also include a fixed-cost component incurred by the SPDs
102 due to the selected mode of operation. For example, the
fixed-cost incurred may be due to being in an active mode (e.g.,
instead of being in a conservation and/or power recovery mode)
during the activity. The determination of the computational costs
may be varied according to the design criteria of a particular
implementation.
The SPDs 102 may be updated with new terminal tasks and/or a new
instruction set. For example, the new terminal tasks may implement
the activities more efficiently and/or safely. In another example,
the new terminal tasks may reduce a time the SPDs 102 are in a
maintenance mode. Updating the SPDs 102 may carry an associated
computational cost in terms of battery.
Use of the resources associated with the SPDs 102 may have a cost.
The remote device(s) may also determine a cost associated with the
SPDs 102. For example, a cost of each of the activities may have an
SPD-component and/or a resource component. The SPD cost component
may assess an intrinsic cost to the SPDs 102 but not the cost of
the resource paired with the SPDs 102 (e.g., the resource cost
component).
In some embodiments, a subset of the activities may be purely
SPD-oriented. The SPD-oriented activities may be necessary to
enable and/or successfully accomplish other activities. For
example, the SPD-oriented activities may comprise sending updates
to the SPDs 102, running diagnostics on the SPDs 102, downloading
data collected during the activities, etc. The type of SPD-oriented
activities may be varied according to the design criteria of a
particular implementation.
The SPDs 102 may be added to the set of resources Q. The SPD
battery may be the resource in the vector of resource capacities R.
The SPD usage may in the resources requirement vector J. The remote
device(s) 104 may be configured to run the refinement rounds to
allocate the resources (e.g., the SPDs 102) and/or determine the
precedence relationships.
The SPD-oriented activities may be scheduled based on a resource
capacity and/or characteristics of the SPDs 102. For example, one
of the SPDs 102 may need to upload data to the remote device(s) 104
more often due to a smaller memory capacity (e.g., the memory 204)
In another example, one of the SPDs 102 may take a longer amount of
time to process information due to a slower processor (e.g., the
processor 202). The characteristics may comprise a processing
speed, a memory capacity, a wireless communication throughput, a
wireless communication transmission distance, a battery capacity,
etc. The resource capacity of the SPDs 102 may be based on the
characteristics of the SPDs 102. The computational costs of
performing the tasks and/or activities may be based on the
characteristics of the SPDs 102. The types of characteristics and
the scheduling of the SPDs 102 due to the characteristics of the
SPDs 102 may be varied according to the design criteria of a
particular implementation.
The SPD-oriented activities may be necessary to enable and/or
successfully accomplish other of the activities in the scheduling
data. The SPD-oriented activities may be added into the precedence
relationships encoded in the graph G=(V, E). For example, a
glucose-monitoring SPD may be scheduled to monitor a patient at
4:00 pm for a period of 8 hours. If the time-slots are minutes, the
activity may be comprised of 32 sequential activities encoded in
the acyclic graph G=(V, E). If, for example, the SPD characteristic
is a limited amount of storage the SPD may have to periodically
upload data. In another example, due to compliance reasons, the SPD
may have to periodically upload data (e.g., to ensure patient
safety). The remote device(s) 104 may enforce, for example, 8
additional SPD-oriented activities instructing the SPD to upload
the collected data into a cloud service (e.g., the remote device(s)
104). The additional activities may be included in the precedence
relationships (e.g., to ensure they are executed hourly during the
glucose-monitoring activity).
For the glucose-monitoring example, for every 15 minute time-slot
the activity cost has an SPD-component. For 24 out of 32
time-slots, the SPD cost component may be the cost of the SPD being
active plus the cost of collecting the data. For 8 out of 32
time-slots, the SPD cost component may increase by the
computational cost to upload the data (e.g., to the remote
device(s) 104). The SPD cost may count against the available
capacity of the SPD. For example, if the SPD cost is greater than
the available capacity of the SPD then the SPD may be flagged as
unavailable.
Some of the tasks performed by the SPDs 102 may be preamble tasks.
Some of the activities may involve preamble tasks to be sent to
and/or executed by the SPDs 102. For example, the preamble task may
be a message sent informing the SPDs 102 of updated scheduling
data. In another example, the preamble task may be a
self-monitoring diagnostics report performed by the SPDs 102. In
some embodiments, preamble tasks may comprise complex tasks for the
SPD to execute before a scheduled activity may be executed. The
preamble tasks performed by the SPDs may be based on the
characteristics of the SPDs (e.g., the memory capacity, the
processing capability, etc.). The type of preamble tasks may be
varied according to the design criteria of a particular
implementation.
The preamble tasks may be designated as an activity to be completed
prior to another designated activity. For example, the
prioritization of the preamble tasks may be encoded in the
precedence relationships determined using the graph G=(V, E).
The fixed-costs incurred by the SPDs 102 during an activity j may
be tallied (e.g., computed) once regardless of the number of
terminal tasks involved. For example, the fixed-cost component may
be zero if no tasks of the SPD are scheduled for the activity j
(e.g., the SPD is inactive for the activity j). The fixed-cost may
be expressed in the RCPSP formulation by the remote device(s) 104
by replacing the resource requirement rjq_i for the SPD involved in
the activity j with r'jq_i=rjq_i+cjq_i*u(rjq_i). A value cjq_i may
be the fixed-cost component incurred by the SPD for being active. A
value u(rjq_i) may be determined by a function with unit value
whenever the variable resource requirement is strictly positive and
zero otherwise.
In some embodiments, the SPDs 102 may be configured to perform
message coalescence. Message coalescence may comprise aggregating
non-urgent tasks. The aggregated non-urgent tasks may be sent
and/or executed closely together instead of being sent and/or
executed piecemeal. The aggregation of the tasks may allow for
energy savings. For example, the energy savings may be due to
communication efficiencies (e.g., reducing an overhead resulting
from sending multiple messages). In another example, the energy
savings may be due to the SPD operating in a power-saving mode
and/or energy recovery mode for longer periods of time.
In some embodiments, the RCPSP performed by the remote device(s)
104 may be configured to take into account message coalescence. For
example, the fixed-costs incurred by the SPDs (e.g., due to being
in a communication mode one or more times) may be incorporating
along with preamble tasks encoded as the activities in the
precedence relationships determined using the graph G=(V, E).
The remote device(s) 104 and/or the SPDs 102 may be configured to
determine which of the tasks to consider urgent and/or which of the
tasks to consider non-urgent. For example, machine learning
techniques may be implemented to determine an urgency of the tasks
based on historical data, location data, personal data, calendar
data, etc. (e.g., context information). The method of determining
the urgency of the task may be varied according to the design
criteria of a particular implementation.
Some of the tasks may be considered critical (e.g.,
mission-critical) tasks. The critical tasks may be potentially
executable by the SPDs 102 at any time (e.g., the critical tasks
may interrupt a performance of a scheduled task). For example, the
critical task may be dialing 911 for emergency services. The
critical task may be triggered at any time whether or not the
critical task is explicitly scheduled.
The remote device(s) 104 may be configured to account for the
critical tasks when prioritizing the activities to determine the
scheduling data. To ensure the SPDs 102 are capable of performing
any of the critical tasks at potentially any time, the SPDs 102 may
be configured to manage a reservoir of charge (e.g., a reserve
charge). The reserve charge may enable any potential execution of
the critical tasks (e.g., ensure performance of the critical
tasks).
In some embodiments, the SPDs 102 may reserve spare capacity in the
battery 214 commensurate to the tasks that are deemed to be
critical. In another example, the SPDs 102 may have a secondary
battery (e.g., a backup power supply) based on the tasks that are
considered to be critical. The reserve charge may be subtracted
(e.g., removed) from the available resource capacity for the SPD
resource in the RCPSP formulation by the remote device(s) 104. For
example, the resource capacity of the SPDs 102 may be reduced based
on the reserve charge and the remote device(s) 104 may perform
refinement rounds to update the scheduling data based on the
resource capacity.
In another embodiment, the critical tasks may be updated. For
example, new critical tasks may be added and/or previously stored
critical tasks may be retired. The reserve charge may be decreased
whenever critical tasks are retired from the mission-critical task
list. The reserve charge may be increased as new tasks are added to
the mission-critical task list. When the reserve charge is
increased and/or decreased the available capacity of the SPDs 102
may be adjusted accordingly. The SPDs 102 may be marked as
unavailable if the current available capacity/charge fails to meet
the reserve charge requirements (e.g., the SPDs 102 are unable to
perform the critical tasks). For example, the reserve charge may
have a pre-determined threshold and if the reserve charge is below
the threshold for performing the critical tasks the SPD 102 may be
marked as unavailable.
Referring to FIG. 23, a method (or process) 1340 is shown. The
method 1340 may determine a task cost for the activity. The method
1340 generally comprises a step (or state) 1342, a step (or state)
1344, a step (or state) 1346, a decision step (or state) 1348, a
step (or state) 1350, a step (or state) 1352, a step (or state)
1354, a step (or state) 1356, a decision step (or state) 1358, a
step (or state) 1360, a step (or state) 1362, and a step (or state)
1364. The step 1342 may be a start step. In the step 1344, one or
more of the SPDs 102 may receive an activity (e.g., from the remote
device(s) 104). The step 1346 may have one or more SPDs 102
determine the terminal tasks to perform to complete the activity.
Next, the method 1340 may move to the decision step 1348.
The decision step 1348 determines if there are more tasks for the
activity. If so, the method 1340 moves to the step 1350. If not,
the method 1340 moves to the step 1352. The step 1350 adds the
computational cost of the task to the cost of the activity. The
step 1352 adds the fixed cost of the active state to the cost of
the activity. Next, the step 1354 begins communication with the
remote device(s) 104. The step 1356 reports the SPD component cost
of the SPD to the remote device(s) 104. Next, the method 1340 may
move to the decision step 1358.
The decision step 1358 determines if there are updated tasks for
the SPD. If so, the method 1340 moves to the step 1360. If not, the
method 1340 moves to the step 1364 which ends the method. The step
1360 updates the SPD with the new terminal tasks. Next, the step
1362 determines the computational costs of the updates, then the
method 1340 moves to the end step 1364.
Referring to FIG. 24, a method (or process) 1380 is shown. The
method 1380 may update the activity cost based on the SPDs 102. The
method 1380 generally comprises a step (or state) 1382, a step (or
state) 1384, a step (or state) 1386, a decision step (or state)
1388, a step (or state) 1390, a step (or state) 1392, a step (or
state) 1394, a step (or state) 1396, a decision step (or state)
1398, a step (or state) 1400, a step (or state) 1402, and a step
(or state) 1404. The step 1382 may be a start step. The step 1384
may schedule an activity. The step 1386 may have one or more of the
SPDs 102 perform tasks associated with the activity. Next, the
method 1380 may move to the decision step 1388.
The decision step 1388 determines if the SPD needs to perform one
or more self-management activities. If so, the method 1380 moves to
the step 1390. If not, the method 1380 moves to the step 1392. The
step 1390 inserts the self-management activities in the precedence
relationships (e.g., determined by the remote device(s) 104). The
step 1392 determines the cost of the SPD activity. Next, the step
1394 adds the cost of the SPD being active to the cost of the SPD
activity. The step 1396 adds the cost of the SPD collecting data to
the cost for the SPD activity. Next, the method 1380 may move to
the decision step 1398.
The decision step 1398 determines if the SPDs 102 uploaded data
during an activity block (e.g., a time-slot of the activity). If
so, the method 1380 moves to the step 1400. If not, the method 1380
moves to the step 1402. The step 1400 adds the cost of uploading
data to the cost of the SPD for the activity. Next, the step 1402
completes the scheduled activity. The step 1404 ends the method
1380.
Referring to FIG. 25, a method (or process) 1420 is shown. The
method 1420 may coalesce data for communication. The method 1420
generally comprises a step (or state) 1422, a step (or state) 1424,
a step (or state) 1426, a decision step (or state) 1428, a decision
step (or state) 1430, a step (or state) 1432, a step (or state)
1434, a step (or state) 1436, a step (or state) 1438, a step (or
state) 1440, and a step (or state) 1442. The step 1422 may be a
start step. The step 1424 may place one or more of the SPDs 102 in
data collection mode. The step 1426 may store the data for sending
(e.g., in the memory 204). Next, the decision step 1428 may
determine if the task is urgent. If not, the method 1420 moves to
the state decision step 1430. If so, the method 1420 moves to the
state 1432.
The decision step 1430 determines if more data is predicted. If
not, the method 1420 moves to the step 1432. If so, the method 1420
moves to the step 1434. The step 1432 begins a communication mode.
Next, the step 1438 sends the stored data. The step 1440 ends the
communication mode. The step 1442 ends the method 1420. The step
1434 waits for more data. Next, the step 1436 combines the new
collected data with the previously stored data. Next, the method
1420 moves back to the decision step 1428.
Referring to FIG. 26, a method (or process) 1480 is shown. The
method 1480 may reserve capacity in the SPDs 102 for critical
tasks. The method 1480 generally comprises a step (or state) 1482,
a decision step (or state) 1484, a step (or state) 1486, a step (or
state) 1488, a step (or state) 1490, a decision step (or state)
1492, a step (or state) 1494, a decision step (or state) 1496, a
step (or state) 1498, a decision step (or state) 1500, a step (or
state) 1502, a step (or state) 1504, a decision step (or state)
1506, and a step (or state) 1508. The step 1482 generally comprises
a start step. The decision step 1484 determines if one or more of
the SPDs 102 stores one or more mission-critical tasks. If not, the
method 1480 moves to the step 1486. If so, the method 1480 moves to
the step 1490. The step 1486 sets the remote device 104 to not
manage the reserve capacity for the SPDs. The method 1480 then
moves to the step 1488, which ends the method 1480. The step 1490
subtracts the reserve capacity from the available SPD capacity.
Next, the method 1480 may move to the decision step 1492.
The decision step 1492 determines if there are new mission-critical
tasks for the SPD. If so, the method 1480 moves to the step 1494.
If not, the method 1480 moves to the decision step 1496. The step
1494 increases the reserve capacity of the SPD. The method 1480
then moves back to the step 1490. The decision step 1496 determines
if a mission-critical task for the SPD needs to be retired. If so,
the method 1480 moves to the step 1498. If not, the method 1480
moves to the decision step 1500. The step 1498 decreases the
reserve capacity of the SPDs, then moves back to the step 1490.
The decision step 1500 determines whether the SPD has enough
capacity for implementing mission-critical tasks. If not, the
method 1480 moves to the step 1502. If so, the method 1480 moves to
the step 1504. The step 1502 marks the SPD as unavailable, and
moves to the end step 1488. The step 1504 performs the scheduled
tasks. Next, the decision step 1506 determines if a
mission-critical task has been triggered. If not, the method 1480
moves back to the step 1504. If so, the method 1480 moves to the
step 1508. The step 1508 interrupts the scheduled tasks to perform
the mission-critical tasks. Next, the method 1480 moves back to the
step 1500.
The remote device(s) 104 may be configured to avoid dividing tasks
among the SPDs 102 (e.g., implement device coalescence). Device
coalescence among the SPDs 102 may be implemented to reduce
overhead costs, reduce the fixed-costs associated with the SPDs 102
being in an active mode and/or to improve efficiency (e.g., similar
to reasons for implementing message coalescence).
The remote device(s) 104 may implement redundant SPDs. The
redundant SPDs may be a subset of the SPDs 102. The redundant SPDs
may be configured to perform the same and/or similar tasks. In some
embodiments, the remote device(s) 104 may build a network of said
redundant SPDs by updating the instruction sets of the redundant
SPDs (e.g., such that the redundant SPDs are configured to have
instruction sets capable of performing the same and/or similar
tasks).
The redundant SPDs may be configured as a network of sensors. In
some embodiments, the remote device(s) 104 may implement the
redundant SPDs with the SPDs 102 having different characteristics.
For example, the redundant SPDs may have different characteristics
such as fixed-costs, computational costs, memory capacity,
processing capability, communication capability, etc. In another
example, a first SPD (e.g., 102-1) may be configured to perform a
task more efficiently and/or with a different computational cost
than a second SPD (e.g., 102-2). In yet another example, the SPD
102-1 may have a different fixed-cost than the SPD 102-2 based on
the selected mode of operation.
For example, a first task (e.g., task A) may only be performed by
the SPD 102-1 and a second task (e.g., task B) may be performed by
either the SPD 102-1 or the SPD 102-2. The SPD 102-2 may perform
task B more efficiently than the SPD 102-1 (e.g., the computational
cost of task B is higher for the SPD 102-1 than for the SPD 102-2).
If the remote device(s) 104 do not consider the fixed-costs of
activating the SPDs 102-1 and 102-2, the refinement rounds may
determine to run the task A on the SPD 102-1 and run task B on the
SPD 102-2. However, if the fixed-cost of activating the SPD 102-2
is higher than the difference in savings (e.g., the computational
cost savings of performing the task B on the SPD 102-2) then the
refinement rounds may schedule both task A and task B on the SPD
102-1 (e.g., the remote device(s) 104 performs device
coalescence).
The activities may be comprised of tasks that may be
executed/performed by any of the SPDs 102. If the remote device(s)
104 determine that there are no fixed-costs and/or communication
overhead (e.g., the fixed-costs are zero) the refinement rounds may
schedule each activity and/or task to the redundant SPDs capable of
executing the tasks the most frugally (e.g., based on the
computational costs). If the fixed-costs are considered, the
refinement rounds may attempt to lump tasks using fewer SPDs unless
the remote device(s) 104 determine the computational cost savings
warrant activating an additional one of the redundant SPDs (e.g.,
depending on the relative fixed-cost of activating the additional
redundant SPD). A second task may be scheduled to be performed
using a single SPD unless the fixed-cost of activating the
additional SPD and/or variable cost (e.g., the computational cost)
for performing the task using the second SPD is determined to be
more efficient (e.g., lower).
For example, the remote device(s) 104 may compare the fixed-costs
and/or computational costs of implementing the tasks and/or
activities using one of the redundant SPDs with the fixed-costs
and/or computational costs of implementing the tasks and/or
activities using multiple SPDs and determine the more efficient
option. The refinement rounds may determine the scheduling data
based on the more efficient option and/or the availability of the
redundant SPDs.
Referring to FIG. 27, a method (or process) 1520 is shown. The
method 1520 may implement device coalescence. The method 1520
generally comprises a step (or state) 1522, a step (or state) 1524,
a step (or state) 1526, a step (or state) 1528, a decision step (or
state) 1530, a decision step (or state) 1532, a step (or state)
1534, a step (or state) 1536, and a step (or state) 1538.
The step 1522 may be a start step. In the step 1524, the remote
device(s) 104 may update the instruction sets for a subset of the
SPDs 102. Next in the step 1526, the subset of the SPDs 102 may be
implemented as the redundant SPDs. The step 1528 may determine the
fixed-cost and/or the computational cost of performing tasks on the
various redundant SPDs (e.g., based on the characteristics of the
redundant SPDs). Next, the method 1520 may move to the decision
step 1530.
The decision step 1530 may determine whether the fixed-costs for
the redundant SPDs are zero (e.g., not considered). If not, the
method 1530 may move to the decision step 1532. If so, the method
1530 may move to the step 1534. The decision step 1532 may
determine whether the fixed-costs are greater than the
computational cost savings (e.g., between performing the tasks on a
single SPD and/or one or more of the redundant SPDs). If not, the
method 1520 may move to the step 1534. If so, the method 1520 may
move to the step 1536. The step 1534 may schedule the tasks for the
redundant SPDs based on the computational costs. Next, the method
1520 ends at the step 1538. The step 1536 may schedule the tasks
among fewer of the redundant SPDs (e.g., perform device
coalescence) based on the fixed-costs and/or the computational
costs. Next, the method 1520 moves to the end step 1538.
While the foregoing description and drawings represent the
preferred embodiments of the present invention, it will be
understood that various additions, modifications and substitutions
may be made therein without departing from the spirit and scope of
the present invention as defined in the accompanying claims. In
particular, it will be clear to those skilled in the art that the
present invention may be embodied in other specific forms,
structures, arrangements, proportions, and with other elements,
materials, and components, without departing from the spirit or
essential characteristics thereof. The presently disclosed
embodiments are therefore to be considered in all respects as
illustrative and not restrictive, the scope of the invention being
indicated by the appended claims, and not limited to the foregoing
description.
The functions performed by the diagrams of FIGS. 12-14 and/or 17-27
may be implemented using one or more of a conventional general
purpose processor, digital computer, microprocessor,
microcontroller, RISC (reduced instruction set computer) processor,
CISC (complex instruction set computer) processor, SIMD (single
instruction multiple data) processor, signal processor, central
processing unit (CPU), arithmetic logic unit (ALU), video digital
signal processor (VDSP) and/or similar computational machines,
programmed according to the teachings of the specification, as will
be apparent to those skilled in the relevant art(s). Appropriate
software, firmware, coding, routines, instructions, opcodes,
microcode, and/or program modules may readily be prepared by
skilled programmers based on the teachings of the disclosure, as
will also be apparent to those skilled in the relevant art(s). The
software is generally executed from a medium or several media by
one or more of the processors of the machine implementation.
The invention may also be implemented by the preparation of ASICs
(application specific integrated circuits), Platform ASICs, FPGAs
(field programmable gate arrays), PLDs (programmable logic
devices), CPLDs (complex programmable logic devices), sea-of-gates,
RFICs (radio frequency integrated circuits), ASSPs (application
specific standard products), one or more monolithic integrated
circuits, one or more chips or die arranged as flip-chip modules
and/or multi-chip modules or by interconnecting an appropriate
network of conventional component circuits, as is described herein,
modifications of which will be readily apparent to those skilled in
the art(s).
The invention thus may also include a computer product which may be
a storage medium or media and/or a transmission medium or media
including instructions which may be used to program a machine to
perform one or more processes or methods in accordance with the
invention. Execution of instructions contained in the computer
product by the machine, along with operations of surrounding
circuitry, may transform input data into one or more files on the
storage medium and/or one or more output signals representative of
a physical object or substance, such as an audio and/or visual
depiction. The storage medium may include, but is not limited to,
any type of disk including floppy disk, hard drive, magnetic disk,
optical disk, CD-ROM, DVD and magneto-optical disks and circuits
such as ROMs (read-only memories), RAMS (random access memories),
EPROMs (erasable programmable ROMs), EEPROMs (electrically erasable
programmable ROMs), UVPROM (ultra-violet erasable programmable
ROMs), Flash memory, magnetic cards, optical cards, and/or any type
of media suitable for storing electronic instructions.
The elements of the invention may form part or all of one or more
devices, units, components, systems, machines and/or apparatuses.
The devices may include, but are not limited to, servers,
workstations, storage array controllers, storage systems, personal
computers, laptop computers, notebook computers, palm computers,
personal digital assistants, portable electronic devices, battery
powered devices, set-top boxes, encoders, decoders, transcoders,
compressors, decompressors, pre-processors, post-processors,
transmitters, receivers, transceivers, cipher circuits, cellular
telephones, digital cameras, positioning and/or navigation systems,
medical equipment, heads-up displays, wireless devices, audio
recording, audio storage and/or audio playback devices, video
recording, video storage and/or video playback devices, game
platforms, peripherals and/or multi-chip modules. Those skilled in
the relevant art(s) would understand that the elements of the
invention may be implemented in other types of devices to meet the
criteria of a particular application.
While the invention has been particularly shown and described with
reference to the preferred embodiments thereof, it will be
understood by those skilled in the art that various changes in form
and details may be made without departing from the scope of the
invention.
* * * * *